Medieval Arithmetic Geometry

Arch. Hist. Exact Sci.
(2013) 67:637705
DOI 10.1007/s00407-013-0121-5
Geometry and arithmetic in the medieval traditions
of Euclids Elements: a view from Book II
Leo Corry
Received: 30 April 2013 / Published online: 18 July 2013
Springer-Verlag Berlin Heidelberg 2013
Abstract This article explores the changing relationships between geometric and
arithmetic ideas in medieval Europe mathematics, as reected via the propositions of
Book II of Euclids Elements. Of particular interest is the way in which some medieval
treatises organically incorporated into the body of arithmetic results that were formu-
lated in Book II and originally conceived in a purely geometric context. Eventually,
in the Campanus version of the Elements these results were reincorporated into the
arithmetic books of the Euclidean treatise. Thus, while most of the Latin versions of
the Elements had duly preserved the purely geometric spirit of Euclids original, the
specic text that played the most prominent role in the initial passage of the Elements
from manuscript to printi.e., Campanus versionfollowed a different approach.
On the one hand, Book II itself continued to appear there as a purely geometric text.
On the other hand, the rst ten results of Book II could now be seen also as possibly
translatable into arithmetic, and in many cases even as inseparably associated with
their arithmetic representation.
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638
2 Euclids Elements: Book II and geometric algebra . . . . . . . . . . . . . 641
2.1 Elements Book II: an overview . . . . . . . . . . . . . . . . . . . . . 641
Dedicated to my dear friend Sabetai Unguru on his 82th birthday.
Communicated by: Len Berggren.
L. Corry (B)
Tel Aviv University, Tel Aviv, Israel
e-mail: [email protected]
1 3
638 L. Corry
2.2 Geometric algebra: an overview . . . . . . . . . . . . . . . . . . . . 646
2.3 Visible and invisible gures . . . . . . . . . . . . . . . . . . . . . . . 649
3 Book II in late antiquity and in Islamic mathematics . . . . . . . . . . . . 651
3.1 Herons commentary of the Elements . . . . . . . . . . . . . . . . . . 652
3.2 Al-Khw arizm and Ab u-K amil . . . . . . . . . . . . . . . . . . . . . 655
3.3 Th abit ibn Qurra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
3.4 Al-Nayrz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661
4 Book II in the early Latin medieval translations of the Elements . . . . . . 663
5 Book II in other medieval texts . . . . . . . . . . . . . . . . . . . . . . . 667
5.1 Abraham Bar-H
.
iyya . . . . . . . . . . . . . . . . . . . . . . . . . . . 667
5.2 Liber Mahameleth . . . . . . . . . . . . . . . . . . . . . . . . . . . . 675
5.3 Fibonacci . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 677
5.4 Jordanus Nemorarius . . . . . . . . . . . . . . . . . . . . . . . . . . 684
5.5 Campanus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689
5.6 Gersonides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693
5.7 Barlaam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 698
6 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 700
1 Introduction
Book II of Euclids Elements raises interesting historical questions concerning its
intended aims and signicance. The book has been accorded a rather singular role
in the recent historiography of Greek mathematics, particularly in the context of the
so-called geometric algebra interpretation. According to this interpretation, Greek
geometry as epitomized in the works of Euclid and Apollonius isat least in its
fundamental aspectsnothing but algebra in disguise.
1
In 1975 Sabetai Unguru published an article in which he emphatically criticized
the geometric algebra interpretation. He claimed that Greek geometry is just that,
geometry, and that any algebraic rendering thereof is anachronistic and historically
misguided (Unguru 1975). Unguru, to be sure, was elaborating on a thesis previously
put forward by Jacob Klein in his classic Greek Mathematical Thought and the Origin
of Algebra (Klein 1968 [19341936]), about a great divide between ancient and mod-
ern mathematics around the basic conceptions of number and of geometry. Ungurus
article ignited a harsh controversy with several well-known mathematicians with an
interest in history, such as Bartel L. van der Waerden (19031996), Hans Freudenthal
(19051990), and Andr Weil (19061998) (van der Waerden 1976; Freudenthal 1977;
Weil 1978). In spite of the bitter debate, however, the controversy quickly receded and
Ungurus view became essentially a mainstream interpretation accepted by most his-
torians. Ungurus criticism has since stood (at least tacitly) in the background of most
of the serious historical research in the eld.
1
Beginning in the late nineteenth century, this view was promoted by prominent scholars such as Paul
Tannery (18431904), Hieronymus Georg Zeuthen (18391930), Sir Thomas Little Heath (18611940),
and Otto Neugebauer (18991990). Book II became a pivotal focus in the elaboration of the details of this
historiographical perspective.
1 3
A view from Book II 639
Still, once we have acknowledged the misleading way in which the geometric
algebra interpretation explains the mathematics of the past by using ideas that were
not present in Euclids or in Apolloniuss time, some additional, interesting questions
arise that call for further historical research. Thus for instance, as an alternative to
geometric algebra, one may ask for a coherent, purely geometric explanation of the
aims and scope of the ideas originally developed in the Greek mathematical texts. This
question has been illuminatingly addressed, for instance, by Ken Saito concerning
Book II and its impact on Apolloniuss Conics (Saito 2004 [1985]), and I will return
to it below.
A different kind of question that arises when one rejects the geometric algebra
interpretation of Greek geometry concerns the origins and historical development of
this very historiographical view. Thus, it is well known that beginning in late antiq-
uity and then throughout history, in editions or commentaries of the Elements, as well
as in other books dealing with related topics, classic geometrical results were var-
iously presented in partial or full arithmetic-algebraic renderings. It is evident that
this mathematical transformation later affected some of the retrospective historical
interpretations of what Euclid had in mind when originally writing his own text. But
what was the precise interplay between the changes at these two levels, mathematical
and historiographical, in the different historical periods? A full exploration of this
longue dure question, starting from Euclid and all the way down to the historians
who vigorously pursued the geometric algebra interpretation in the late nineteenth
century, is a heavy scholarly task. In the present article, I intend to address a partial
aspect of it, by focusing on the ways in which mathematicians in Medieval Europe
presented the propositions of Book II in the most widely circulated Euclidean ver-
sions, as well as in other texts that incorporated some of the propositions of that book.
I suggest exploring the extent to which arithmetic and proto-algebraic ideas were
absorbed in those texts and hence modied the original Euclidean formulation, and
(to a lesser degree) whether and how these changes affected the historical conception
of Euclid.
Before entering the discussion, however, it is important to stress from the outset
that I deliberately ignore the debate about the adequacy of using the term algebra
in this or that historical context, and consider it a matter of taste. As I have explained
elsewhere, the question about the essence of algebra as an ahistorical category seems
to me an ill-posed and uninteresting one (Corry 2004, 397). Thus, I am not interested
in adjudicating the question whether or not certain specic ideas found in Heron or
in al-Khw arizm count as algebra according to some predetermined, clearly agreed
criteria. Rather, I want to identify those mathematical ideas not originally found in
Euclids text and that were gradually incorporated into interpretations of it or even
into the edited versions of the text itself. In this article, just for convenience, I refer
to some of those ideas using the general umbrella terms of arithmetic or alge-
bra, in their more or less agreed sense, and without thereby aiming at an essentialist
perspective on these concepts. This in itself should not give rise to any debate or
confusion.
The core of the article is preceded by two relatively lengthy introductory sections.
In Sect. 3, I discuss some of the propositions of Book II and their proofs as they appear
1 3
640 L. Corry
in the original Euclidean text known to us nowadays.
2
I devote particular attention to
II.5 which in most of the examples analyzed here serves well as a focal point that
illustrates the issues considered.
3
I thus present the typical (anachronistic) algebraic
interpretation of this result and I discuss its shortcomings. In Sect. 4, I present some
versions of the same propositions, as were introduced in texts of late antiquity and of
Islam mathematics. This section is not intended as an exhaustive survey of such texts,
but rather as a presentation of versions that were available to the medieval translators
and interpreters of Euclid and that in many cases provided the source for contemporary
renderings of propositions of Book II. These two sections, Sects. 34, are intended
as a consistent synthesis which, while strongly relying on the existing scholarship,
stresses some less noticed aspects and thus hopefully helps making a fresh reading of
the history of Book II in its early phases.
After these introductory sections, I move to the core part of the article, where I ana-
lyze propositions from Book II as they appear in medieval versions of the Elements
(in Sect. 5) and in other contemporary books that incorporated such ideas (in Sect. 6).
The period discussed comprises the rst Latin translations of the Euclidean text in
the twelfth century, and it also includes additional texts that circulated in Europe in
manuscript versions before the rst printed version of the Elements in 1482. Thus,
I discuss the Liber Mahameleth, as well as additional texts by Abraham Bar-H
.
iyya,
Fibonacci, Jordanus Nemorarius, Gersonides, and Barlaam. These works present us
with a rather heterogeneous variety of approaches, within which propositions from
Book II were handled with the help of both geometric and arithmetic ideas. Some of
these treatises organically incorporated into the body of arithmetic the main results
of Book II. Subsequently, and particularly in the Campanus version of the Elements,
existing arithmetic versions of results fromBook II were reincorporated into the arith-
metic books the Euclidean treatise, Books VIIIX. As a consequence, while most of
the Latin versions of the Elements had duly preserved the purely geometric spirit of
Euclids original, the specic text that played the most prominent role in the initial
passage of the Elements frommanuscript to printi.e., Campanus versionfollowed
a different approach. On the one hand, Book II itself continued to appear as a purely
geometric text. On the other hand, the rst ten results of Book II could now be seen
also as possibly translatable into arithmetic, and in many cases even as inseparably
associated with their arithmetic representation. Hence, when symbolic techniques of
algebra started to take center stage in renaissance mathematics in Europe, the alge-
braic interpretation of results in Book II could became a natural, additional step to be
followed.
In the discussion below, as already indicated, proposition II.5 provides a signicant
focal point for the analysis pursued. Nevertheless, this focus allows only a limited
2
The issue of the adequate use of direct and indirect sources in order to establish an authoritative version of
the Euclidean text is a signicant historiographical question still under debate nowadays. See (Rommevaux
et al. 2001). I do not deal at all with this issue here, and I simply follow the widely accepted English version
(Heath 1956 [1908]).
3
Proposition II.5 has also been used as focal point in other accounts about Greek mathematics or about
changing views on geometry through history; see, e.g., Neal 2002, 123 ff.; Netz 1999, 911; Vitrac and
Caveing 1990, 370372.
1 3
view on the broader issue at stake. On closer analysis, one realizes that another issue
of crucial importance in this story is the way in which distributivity of the product
over addition was handled in the various medieval texts that I study here. In consid-
eration with the already substantial length of this article, I have pursued that issue in
a separate essay, soon to be published, and to which I refer in the relevant places as
[LC2].
Another remark related to the length of my article is the following: Since, of neces-
sity, the text belowincludes many detailed proofs that differ fromeach other in specic
important points, yet in subtle and perhaps non-dramatic manners, I have followed the
convention of writing some paragraphs using a different font. This is meant
as an indication to the reader that these paragraphs comprise purely technical descrip-
tions of proofs and that they are intended as evidence in support of the general claims
made in the corresponding sections. The paragraphs may be read with due technical
attention, or they may be skipped at least temporarily without thereby missing the
general line of argumentation.
2 Euclids Elements: Book II and geometric algebra
Book II of the Elements is a brief collection of only fourteen propositions. The rst
ten can be seen as providing relatively simple tools to be used as auxiliary lemmas in
specic, more complex constructions later on. Euclid applied the results in the proofs
of the last four propositions of Book II (II.11, for instance, teaches how to obtain the
mean and extreme section of any given segment), as well as in other books of the
Elements. Later on, they can be found in important places such as Apollonius Conica.
Each of the rst ten propositions was proved by Euclid directly with the help of results
taken from Book I, and without relying on any other result from Book II. It is easy to
see, however, that once II.1 is proved on the basis of I.34, the other nine propositions
could be proved by relying on II.1. This is not what Euclid did in his text, but as we
shall see, the possibility was acknowledged from very early on.
2.1 Elements Book II: an overview
Since I will be referring throughout the article to some propositions in Book II and to
their proofs, I present now propositions II.1II.6 and II.9II.10, with varying degrees
of detail (according to what is needed in the discussion below). I also add some general
comments at the end of the presentation (all the quotations are taken from Heath 1956
[1908]):
4
II.1: If there are two straight lines, and one of them is cut into any number of
segments whatever, then the rectangle contained by the two straight lines equals
4
Recent scholarship has devoted increased attention to the ways in which diagrams appearing in critical
editions of the Elements differ fromthose in extant manuscripts. See, e.g., (Saito and Sidoli 2012). Although
considerations of this kind may be relevant to our analysis here, in this article I will refer only to diagrams
as they appear in the available critical editions.
1 3
642 L. Corry
the sum of the rectangles contained by the uncut straight line and each of the
segments.
Figure 1
Here A is the first given straight line, and BC is the second one. BC is
divided into three parts: BD, DE, EC (this is meant to indicate that
BC is cut into any number of segments whatever). The construction is
straightforward, setting BG equal to A, and perpendicular to BC, and
then drawing DK, EL, and CH parallel to BG. All that is necessary to
complete the proof is that DK be equal to A. This follows from the fact
that BK is by construction a rectangle, and hence (by I.34) DK, being
opposite to BG, equals BG. For the same reason also EL and CH are equal
to BG, and all the three parallelograms can be taken together to form
the parallelogram BH.
II.2: If a straight line is cut at random, then the sum of the rectangles contained
by the whole and each of the segments equals the square on the whole.
Figure 2
II.3: If a straight line is cut at random, then the rectangle contained by the whole
and one of the segments equals the sumof the rectangle contained by the segments
and the square on the aforesaid segment.
1 3
Figure 3
II.4: If a line is cut at random, then the square on the whole is equal to the squares
on the segments and twice the rectangle contained by the segments.
Figure 4
II.5: If a straight line be cut into equal and unequal segments, the rectangle con-
tained by the unequal segments of the whole together with the square on the straight
line between the points of section is equal to the square on the half.
The straight line AB is cut into equal segments at C and into unequal segments at
D. The proposition says that the rectangle on AD, DB together with the square on CD
is equal to the square on CB. The diagram is built by taking AK = AD, BF = CB.
Figure 5
1 3
644 L. Corry
In the proof Euclid makes use of a gnomon, NOP, which is the gure obtained
when joining the rectangles HF, CH, together with the square DM. Euclids proof can
be visualized as follows:
Figure 6
Schematically, Euclids argument can be summarized as follows
[Sq(CD) means the square of CD and R(CD, DH) means the rectangle
built on CD, DH]:
(a.1) By I.43: R(CD,DH) =R(HM,MF), and hence R(CB,BM)=R(DB,BF)
(a.2) But R(CB,BM)=R(AC,AK) and hence R(AC,AK)=R(DB,BF)
(a.3) Hence R(AC, AK)+R(CD,DH) =R(DB,DF)+R(CD,DH), or,
(a.4) R(AD,AK) =Gnomon NOP or R(AD,DB) =Gnomon NOP
(a.5) Hence R(AD,DB) +Sq(LH) =Gnomon NOP +Sq(LH) or,
(a.6) R(AD,DB) +Sq(CD) =Sq(CB)
II.6: If a straight line is bisected and a straight line is added to it in a straight
line, then the rectangle contained by the whole with the added straight line and the
added straight line together with the square on the half equals the square on the
straight line made up of the half and the added straight line.
1 3
The segment AB is bisected at C, and a segment BD is added to it in a straight line.
The proposition says that the rectangle AD, DB together with the square on CB equals
the square on CD. The diagram is built by taking AK = BD, DF = CD.
Figure 7
Like in II.5, also in this case, a key step in Euclids proof is based on the use of a
gnomon, NOP, and of II.43 in order to assert that CH = HF.
II.9: If a straight line be cut into equal and unequal segments, the squares on the
unequal segments of the whole are double of the square on the half and of the
square on the straight line between the points of the section.
The straight line AB is cut into equal segments at C and into unequal segments at
D. The proposition says that the squares on AD, DB are double of the squares on AC,
CD.
Figure 8
The proof refers to the diagram above where GF = CD, EC = AC = CB, and
wherein the angles AEB, ECA, ECB are easily shown to be right angles.
The Pythagorean theorem (I.47) is thus applied several times, as
follows:
(b.1) By construction and by I.47: Sq(AE) =2 SQ(AC)
(b.2) By construction and by I.47: Sq(EF) =2 SQ(GF) =2 SQ(CD)
(b.3) Hence: Sq(AE) +Sq(EF) =2 SQ(AC) +2 SQ(CD)
1 3
646 L. Corry
(b.4) But by I.47: Sq(AF) =Sq (AE) +Sq(EF)
(b.5) Hence [from (b.3) and (b.4)]: Sq(AF) =2 SQ (AC) +2 SQ(CD)
(b.6) But by I.47: Sq(AF) =Sq(AD) +Sq(DF) =Sq(AD) +Sq(DB)
(b.7) From (b.5) and (b.6): Sq(AD) +Sq(DB) =2 Sq(AC) +2 Sq(CD)
II.10: If a straight line is bisected, and a straight line is added to it in a straight

line, then the square on the whole with the added straight line and the square on
the added straight line both together are double the sum of the square on the half
and the square described on the straight line made up of the half and the added
straight line as on one straight line.
The segment AB is bisected at C, and a segment BD is added to it in a straight
line. The proposition says that the squares on AD, DB are double of the squares on
AC, CD.
Figure 9
2.2 Geometric algebra: an overview
Even a supercial glance at II.1 immediately indicates why the algebraic interpretation
readily suggests itself to a modern reader, who would identify here a particular case
of the distributive property of multiplication over addition seen as an algebraic rule.
Of course, this identication requires that we add some ideas that do not appear in the
text as cited, such as equating area formation with multiplication of abstract quantities.
But a reader who is willing to ignore the historical context will have no difculty in
doing so, even though what the texts displays is a purely geometric formulation and
proof of a property of area formation.
The situation becomes somewhat more complex and interesting if we look at II.5
and II.6. In his well-known comments to these propositions, Heath gave the typi-
cal geometric algebra interpretation, by assigning algebraic symbols to segments in
the diagram of II.5, as follows: AD = a, BD = b, hence CB = (a + b)/2 and
1 3
CD = (a b)/2 (Heath 1956 [1908], Vol. 1, 383). In these terms, the proposition
can be interpreted as an algebraic identity, namely,
a +b
2
a b
2
2
= ab.
One interesting feature of his interpretation is that the relevant expression for the
proposition changes if we call the segments differently. In particular, the two propo-
sitions, II.5II.6, can be made to represent one and the same expression, namely the
difference of the squares of two straight lines is equal to the rectangle contained by
their sum and difference, or (a +b) (a b) = a
2
b
2
. This equivalent formulation
obtains if the two lines in Fig. 10 below represent the two propositions, with a and b
being CB and CD, respectively, whereas AD and BD are taken as their sum and their
difference respectively. Thus:
Figure 10
Moreover, the two same lines could be taken to represent a putative algebraic
interpretation common to II.9II.10: The sumof the squares on the sumand difference
of two given straight lines is equal to twice the sum of the squares on the lines or
(a +b)
2
+(a b)
2
= 2 (a
2
+b
2
) (Heath 1956 [1908], Vol. 1, 394). This interpretive
exibility may be seen as either an advantage or a disadvantage, depending on what
viewpoint we adopt, but it certainly raises some concerns that require being clearly
addressed.
Thus, if we remain close to the Euclidean text we have to admit that, particularly in
the cases of II.5 and II.6, both the proposition and its proof are formulated in purely
geometric terms. There are no arithmetic operations involved, and surely there is no
algebraic manipulation of symbols representing the magnitudes involved. The entire
deduction relies on the basic properties of the gures that arise in the initial construc-
tion or that were proved in previous theorems (which in turn were proved in purely
geometric terms). Thus, for instance, the claim that the complement CH is equal to
the complement HF corresponds to proposition I.43 of the Elements. The gnomon
NOP is a geometric gure built out of other gures, and similar gnomons appear in
many other proofs in Greek geometry. It is clear that, while one might easily claim
(albeit with little historical justication, but at least with some mathematical justica-
tion) that rectangle formation is a geometric equivalent of arithmetic multiplication,
no such natural, arithmetic equivalent can be suggested for gnomon formation.
Ungurus criticism of the geometric algebra interpretation laid stress on this kind of
interpretive difculties concerning the meaning of the operations. It also indicated the
inherent difculty to dene a clear, general arithmetic of abstract magnitudes in Greek
1 3
648 L. Corry
mathematics, of which the putative algebra would be a generalization (Unguru and
Rowe 19811982a, b; see also Mueller 1981, 5052).
Whether or not one accepts this kind of criticism, it is pertinent to notice that when
it comes to the arithmetical books of the Elements, Books VIIIX, the discussion may
require some specic adaptations, since the arguments against the algebraic interpre-
tation of geometric situations cannot simply be extended without further comments.
I discuss this issue more at length in [LC2]. Nevertheless, I want to stress here one
point concerning the way in which numbers and operations upon themare represented
in the arithmetical books of Euclid, while contrasting it with the case of the geometric
books. This is a very important point for our discussion below on the medieval texts.
The accompanying diagram for VII.5 is useful for our purposes here:
Figure 11
As in all diagrams associated with the arithmetical propositions in the Elements,
lines represent here numbers while the operations of addition and subtraction can be
represented by concatenation of lines. In this diagram, for instance, EF is the sum of
EH, HF. This is not different fromaddition of two lines in the geometric context. When
it comes to multiplication, however, we need to be more attentive to the differences
between the two contexts. The number A, for instance, is a part of BC, which means
that A, if added repeatedly to itself, yields BC. In the proof, Euclid simply counts and
compares the multitude of times that one number is part of the other in the cases
considered. I will be referring below to this kind of argument as counting units.
In the Euclidean texts, such arguments appear in the context of arithmetic proofs,
where lines appearing in diagrams are used just to represent and to label arbitrary
numbers and actual constructions are never performed on them. These lines are not
multiplied with each other, but they can be multiplied in the sense of repeated addition.
In contrast, in geometric proofs two lines may be multiplied in the sense of rec-
tangle formation. Whereas in the framework of the Elements, the separation is clearly
kept between the geometric and the arithmetic books, and the kinds of proofs used
in each, in the framework of Islam mathematics, and certainly later on in medieval
European mathematics, we nd proofs of both types mixed together in various con-
texts.
5
Among the interesting changes that we shall notice below is that arithmetic
proofs started to be increasingly used also for propositions originally appearing in
Book II.
5
The distinction between the two kinds of proofs has efciently been used in (Oaks 2011) for the case of
Islam mathematics.
1 3
2.3 Visible and invisible gures
A possible reaction to any criticism against the geometric algebra interpretation of
Book II is to ask for a coherent, purely geometric interpretation of the meaning and
usage of this collection of propositions seen as a whole. One illuminating such inter-
pretation has been suggested by Saito (2004 [1985]), and it is also relevant for our
discussion below. Saitos interpretation is related to the twin-like relation already
mentioned above between pairs of propositions such as II.5II.6 or II.9II.10. Simi-
lar relations can be shown to hold, respectively, for the pairs II.2II.3, II.4II.7, and
II.1II.8. We saw that in the algebraic interpretation, these pairs of twin propositions
may be understood as representing one and the same algebraic expression. But this
immediately raises the question why would the same expression require two different
propositions to express it. This would seem to go against the gist of the algebraic
perspective as known to us nowadays, in which precisely such repetitions become
unnecessary. One of Saitos interesting insights is that the geometric context alone
provides a very coherent and sufcient explanation for the existence of these twin
propositions that obviates the need for an algebraic addition (Saito 2004 [1985], 157
160). Indeed, Saito points to several places in the Elements, in other books of Euclid,
and in Apollonius Conica, where two different, but strongly related, geometric sit-
uations are proved with the help of twin propositions from Book II. An example of
this appears in propositions III.35III.36 of the Elements, which deal with areas of
rectangles built on segments of lines that intersect with each other and with a circle.
In III.35, the lines intersect within the circle, whereas in III.36 they intersect outside
it. Two diagrams appearing in the proofs of the propositions and related to this point
are the following:
Figure 12
1 3
650 L. Corry
In III.35, the line AC is cut into equal parts at G and into unequal ones on E, and thus
II.5 can be applied. In III.36, the line AC is bisected at F and DC is added to it, and thus
II.6 can be applied. Notice then, that not only the relations between the lengths here
are important, but above all their geometric arrangement. Thus, Saito concludes that
Euclid considered lines and areas not as representations of abstract quantities that can
be freely manipulated according to general rules, but specically as geometric entities,
the mutual arrangement of which is signicant for the propositions considered.
Saitos analysis shows that a purely geometric interpretation of Book II does full
justice to Euclid as a conscious planner of the mathematical edice of the Elements. In
later books of the collection, we encounter geometric situations that require the support
of lemmata such as those put forward in advance in Book II in order to complete the
proof or the construction at stake. The very singling out of these results as worthy
of separate consideration in advance appears in retrospect as an under-acknowledged
token of Euclids great insight. I say this, because these lemmata came to be used
signicantly not only where originally intended, but also (as Saito shows) in Euclids
Data and, somewhat later, in Apolloniuss Conica. And as the mathematician Doron
Zeilberger emphasized recently, A Good Lemma is Worth a Thousand Theorems,
precisely because, while trivial in appearance or easy to prove, once stated, the insight
encapsulated in a good lemma allows for its application in a wide variety of unexpected
contexts, and this was indeed the case with Book II.
6
Another aspect of Saitos analysis concerns the distinction between visible and
invisible gures in the diagrams of the Elements. This important point is also connected
withthe issue of distributivitywhichI discuss infurther details in[LC2]. Here I mention
it only briey. Recall that the geometric fact to be proved in II.1 can be schematically
stated as follows (referring to Fig. 1 above):
R(A, BC) = R(A, BD) +R(A, DE) +R(A, EC) . (1)
The proof itself, on the other hand, is based on (i) taking a segment BG = A, (ii)
constructing the parallelograms and proving on purely geometric grounds (using I.34)
that DK = A = EL, and (iii) then realizing that, according to the diagram:
R(BG, BC) = R(BG, BD) +R(DK, DE) +R(EL, EC) . (2)
So, what is the big difference between (Eq. 1) and (Eq. 2) and in what sense does
the latter prove the former? Notice, in the rst place, that proving DK = A = EL is
fundamental since otherwise the three rectangles in the gure cannot be concatenated
into a single one in (Eq. 2). But what Saito draws our attention to, in particular, is
the fact that the rectangles used in (Eq. 2) are all visible in the diagram, whereas
those of (Eq. 1) are invisible. Situations such as those of (Eq. 2) appear frequently
in the Elements, and the distributivity of the construction of parallelograms is used
6
See Doron Zeilberger, Opinion 82: A Good Lemma is Worth a Thousand Theorems (written: August
14, 2007; downloaded May 02, 2012: http://www.math.rutgers.edu/~zeilberg/Opinion82.html): Theorems
are nice, but they are usually dead-ends. A lemma may be trivial, or easy to prove once stated, but if it is
good, its value far surpasses even the deepest theorems.
1 3
there without any further comment. A most prominent example appears in the proof
I.47, whose well-known diagram is the following:
Figure 13
A crucial step in the proof is that
Sq (BC) = R(BD, DL) +R(CE, LE) , (3)
and this step is taken in Euclids text without any special comment. In other words, sit-
uations embodied in (Eq. 2) and (Eq. 3) involve visible gures and hence do not require
further justication other than what the gure itself shows. The situation embodied in
(Eq. 1), in contrast, does require a proof precisely because the rectangles involved are,
as indicated by Saito, invisible. In Book II, then, Euclid shows how the properties of
invisible gures can be derived from those of visible ones for one can apply to the
latter the geometric intuition which is fundamental of Greek geometric arguments
(Saito 2004 [1985], 167). In the texts discussed below, the awareness to this clear dis-
tinction is not strictly kept, and the blurring of borders between the two kinds of gures
runs parallel to the processes of blurring of borders between geometric, arithmetic,
and proto-algebraic ideas.
3 Book II in late antiquity and in Islamic mathematics
Medieval readers, translators, and editors of the Elements were acquainted with various
kinds of commentaries and additions, and not just with the original Euclidean text such
1 3
652 L. Corry
as described above for the particular case of Book II. In this section, I present some
versions of results related to Book II, written in late antiquity and in the mathematical
culture of Islam, and that circulated in Europe since the twelfth century. The texts
discussed here had a direct impact on the way that the results of Book II came to be
interpreted, reproduced, and disseminated in mathematical texts in the middle ages.
3.1 Herons commentary of the Elements
Alternatives to Euclids proofs started to appear already within Greek mathematical
culture itself. It has been speculated that arithmetical versions of some propositions
in Book II circulated at the time of Diophantus and perhaps even earlier (Vitrac 2004,
22). Also, a number of scholia to the Elements include an arithmetic rendering of
II.5, with values AB = 10, AC = CB = 5, AD = 8, and DB = 2 (Heiberg
and Menge 18831893, Vol. 5, 234236), but their exact dating and authorship is a
debatable matter (Vitrac 2003). What can be asserted with relative certainty is that
Heron of Alexandria, who at the end of the rst century A.D. wrote a Commentary
of the Elements, presented original, alternative proofs to many propositions in Book
II. These proofs provide a good illustration of how Euclids arguments started to be
transformed from quite early on. Since Herons text was known, either directly or
indirectly, to some of the medieval authors about whom we shall speak below, his
proofs are worthy of examination here.
Reconstructing the exact contents of Herons text is not a straightforward task, since
only very meager fragments have survived in Greek (Vitrac 2011). The main available
substantial source for the existing reconstructions is found in a commentary to the
Elements written by Abul Abbas al-Fadl ibn Hatim al-Nayrz (c. 875c. 940). This
commentary (which I discuss below) was one of the earliest to be written in Arabic,
and it preserved a considerable number of extracts from Herons book (Heath 1981
[1921], 309310). The medieval authors discussed in this article became acquainted
with Herons ideas via al-Nayrzs commentary, and hence, it seems reasonable to
rely on it for our discussion here. That being said, it is nonetheless important to keep
in mind that more recent historical research has stressed the difculties in asserting
the ways in which the extant Latin and Arab manuscripts reect the original text of
Heron (Busard 1996b; Brentjes 2001a).
As already mentioned, Euclid had proved each proposition in Book II separately on
the basis of results of Book I alone. Heron followed a different approach. He asserted
that II.1 is the only one among the fourteen propositions that cannot be proved without
drawing a total of two lines. As for the remaining propositions, however, he stated
that it is possible that they be demonstrated with the drawing of one sole line, and
he suggested alternative proofs that do not rely anymore on Book I. Rather, he relied
in each case on propositions from the same Book II that he gradually proved as he
went along (Curtze 1899, 8889). Thus, II.1 appears here as the basic statement of
a general law of distributivity of area formation over addition, a law whose proof is
purely geometric, and from which all other propositions in Book II can be derived.
Propositions II.2II.3 appear as particular cases of II.1, and II.4 as directly derivable
from it (see [LC2]). Hence, implicitly, also these propositions derive their validity
1 3
from geometry, but at the same time, they embody situations which Heron saw as
arithmetic and illustrated with numerical examples. Along the proof, he also referred
to rectangles and squares constructed on the various segments (or, as he phrased it,
the surface that the two lines CD,DB enclose), but such gures are truly invisible,
in the sense of Saito, i.e., they are never actually drawn and it is left to the reader to
imagine them. In addition, Heron stressed that each proposition can be proved in two
different ways, namely by analysis and by synthesis. I will present now the details of
Herons proof of II.5, while focusing only on the second of these two components,
namely synthesis.
Herons proof for II.5 relies directly on II.2II.3, rather than on
propositions in Book I, as in Euclids original. Heron draws a line
AB, with two additional points D and C indicated on it, and with C
bisecting the line, as follows (Curtze 1899, 96):
7
Figure 14
The core of the proof can be visualized in terms of two simple steps:
(1) decomposing the square on BC into smaller pieces; (2) reassem-
bling the pieces into the rectangle on DA,DB and the smaller square
on DC. Graphically this amount to the following (I am using figures
that do not appear in the text):
Figure 15
Schematically, Herons argument can be summarized as follows:
(c.1) By II.2: Sq(CB) =R(CB,DB) +R(CB,CD)
(c.2) But, by II.3 R(CB,CD) =R(CD,DB) +Sq(CD) (since BC =DB +CD)
(c.3) Hence Sq(CB) =R(CB,DB) +(R(CD,DB) +Sq(CD))
(c.4) But AC =CB
(c.5) Hence Sq(CB) =R(AC,BD) +(R(CD,DB) +Sq(CD))
(c.6) But by II.1 R(AC,DB) +R(DB,CD) =R(AD,DB)
(c.7) Hence Sq(CB) =R(AD,DB) +Sq(CD),
Historians have identied Herons proofs as early instances of using algebraic
techniques in geometry, and this assessment remained unchallenged even in critical
7
And for an English translation, see (Lo Bello 2009, 32).
1 3
654 L. Corry
analyses of geometric algebra (see, e.g., Fried and Unguru 2001, 2021). This way
to interpret Herons work, however, seems to me misleading, as I nd it hard to see this
approach as algebraic in any possible sense of this word. In the proof just presented,
as well as in others in Book II, Heron added areas to areas or decomposed a square into
geometric components, and then manipulated the parts in order to reconstruct a differ-
ent one. On the one hand, these are just legitimate geometric operations also found in
other parts of the original Euclidean text. On the other hand, there is here an interesting
twist of ideas whereby Heron extended the scope of Euclids norms, in the sense that
he applied to invisible gures manipulations that Euclid had legitimately applied only
to visible ones. Indeed, as we saw above, the whole idea of Book II was to provide
tools that created a sound basis for doing these kinds of geometric manipulations
wherever needed, and now we nd here the manipulation of invisible gures being
done already within Book II. Moreover, one cannot overlook the important difference
embodied in the fact that Euclids proof is in essence constructive, while Herons is
operational: what I mean by this is that Euclid starts with an elaborate construction
that needs to be completed before starting the apodictic part of the proof, whereas
Heron proceeds straightforwardly from the divided line to the conclusion, simply by
operating (i.e., adding and comparing areas) with squares and rectangles built on the
segments appearing in the proposition (and using some previous propositions as well).
In fact, it is quite clear that Herons diagram alludes to those appearing in the arith-
metical books of the Elements, which typically comprise just collections of segments
which are referred to in the proof but are not used for any kind of construction. Thus,
Herons proof are different from Euclids in important senses, but not in the sense of
being algebraic rather geometric. Indeed, they are not even arithmetic proofs in the
sense explained above, since two lines multiplied give raise to a rectangle and not to
a third line. Herons proofs are not less geometric than Euclids, but rather differently
geometric, and in a meaningful manner at that.
8
Given this different geometric approach, some have speculated about the possi-
bility that Herons proof had its sources in Pythagorean ideas from a time in which
arithmetic and geometric practices were less clearly separated than what they came
to be later on, as in the Euclidean text (Vitrac and Caveing 1990, 369). At any rate,
it seems evident that the operational character of Herons proof, even though it is
geometrically operational, more conveniently prepares the road for a possible arith-
metic, and later algebraic, readings of the propositions in Book II. As we shall see
below, this road was indeed taken by later readers of Euclids text and of Herons
commentary. But it is important to stress these kinds of differences between two
geometric approaches (Euclids and Herons) since it is precisely through this kind
of nuances that we come to understand the slow process through which algebraic-
like thinking entered geometry and in particular the kind of geometry developed in
Book II.
8
(Vitrac 2005, 6 ff.) speaks about two contrasting styles of geometrical proof in Greek mathematics:
demonstrative versus algorithmic. Without wanting to make too much of word choice, and without
the benet of the much broader scope of Vitracs analysis, I think that for the case of Book II, at least,
the contraposition of constructive versus operational is more adequate to encapsulate the difference
between Euclids and Herons proofs.
1 3
3.2 Al-Khw arizm and Ab u-K amil
The development of procedures for solving problems involving unknowns and their
powers was a central contribution of the mathematicians in the culture of Islam. The
terms Islamic algebra or Arabic algebra can be associated with this tradition.
Their most well-known contributions are dated not before the beginning of the ninth
century, but it is likely that some of the earlier ideas began as practical traditions
that were cultivated and transmitted orally over many centuries (Hyrup 1986). It is
obvious, at any rate, that the Islamic context must be explored in any attempt to follow
the incorporation of arithmetic, algebraic or proto-algebraic ideas into later versions
of Book II. One must keep in mind, of course, that the transmission of the Arabic
Elements involved a highly complex network of translations, editions, commentaries,
and reception, about which knowledge continues to be somewhat limited and contested
(Brentjes 1994, 1996, 2001a). The current scholarship typically refers to two basic
Arabic translations of Books IXIII that gave rise to separate textual traditions. One,
composed before 805, is attributed to Al-H
.
ajj aj ibn Y usuf ibn Mat
.
ar (. between 786
833). A second one, composed by Ish
.
aq ibn H
.
unayn (ca. 830910/11), during the last
third of the ninth century, was later edited by Th abit ibn Qurra (ca. 830901). I will
discuss here ideas of Book II that appear in the works of al-Khw arizm, Ab u K amil,
Th abit ibn Qurra, and al-Nayrz. In all these cases, these ideas appear in contexts
that sensibly differ from the original Euclidean one. Moreover, in each case we nd
different approaches to the way in which the result can be used and interpreted from
an algebraic or arithmetic perspective. These four mathematicians do not exhaust the
variety of relevant texts from the Islamic tradition, but they were among the most
commonly read in medieval Europe, and this is the reason for focusing here on their
works.
The famous Al-kit ab al-muh
tas
.
ar f h
.
is ab al-jabr wa-l-muq abala (The Com-
pendious Book on Calculation by Restoration and Confrontation) was written by
Muh
.
ammad ibn M us a al-Khw arizm (c. 780850) in the early ninth century on the
exhortation of the caliph al-Mamun. As it is well known, al-Khw arizm presented here
rules for solving problems that involve squares of an unknown quantity, and he then
added geometric proofs to justify some of the rules. Although it seems unlikely that he
was not aware of some of the existing translations of Euclids text, the fact is that he
did never directly refer to or otherwise mention the Elements (Djebbar 2005, 3436).
Neither did he explicitly state or prove results of Book II in any of his works. Many
of the geometric proofs found in his texts bear similarities with the Euclidean propo-
sitions, but they are used in a less rigorous and more intuitive or visual manner than
in the original. A well-known example is the use of a result similar to II.5 in relation
to the problem known as the square and twenty-one numbers equal ten roots of the
same square. Here, we nd an early, interesting case of embedding the core of II.5
in a typical Arabic algebraic context.
Al-Khwarizms procedure to solve this problem is described as fol-
lows (Rosen 1831, 11):
Halve the number of the roots; the moiety is five. Multiply this by
itself; the product is twenty-five. Subtract from this the twenty-one
which are connected with the square; the remainder is four. Extract
1 3
656 L. Corry
its root; it is two. Subtract this from the moiety of the roots, which
is five; the remainder is three. This is the root of the square which
you required, and the square is nine. Or you may add the root to the
moiety of the roots; the sum is seven; this is the root of the square
which you sought for, and the square itself is forty-nine.
The diagram used to endorse the validity of this procedure is remi-
niscent but not identical to the Euclidean one for II.5. It involves
a square AD, whose side AC represents the unknown magnitude. In the
diagram there is also a rectangle HT, one of whose sides, HN, equals
AC.
Figure 16
The diagram accounts for the problem in the sense that rectangle HB
and square AD together build a larger rectangle, HD, that represents
10 times the unknown magnitude AC, while HB one is assigned the value
21. Al-Khwarizms argument, from here on, can be schematically ren-
dered as follows:
(d.1) Bisect HC at G, and construct square MT with side equal to
HG. Since HG is of length 5, then the area of MT is 25.
(d.2) Construct KMHG with KG =GA. Here KR =Sq(GA).
(d.3) Now, we have cut HC into equal segments at G and into unequal
segments at A. Euclids II.5 can be applied here, so that:
Sq(HG) =R(HA,AC) +Sq(GA). Hence, MT =HB +KR.
(d.4) Thus, the value of KR is 4, and its side is 2. And since
GK =GA, it follows that AC is 3, and this is the side we were
looking for.
Al-Khw arizm is thus using here the main idea behind II.5 in the framework of a
specically arithmetic case. He freely associates numerical values to what for Euclid
are continuous magnitudes (line segments), andthenhe canobtain, withthe helpof II.5,
another value that is associated to a certain square. Of course, this association crucially
depends on the conceptions of number typical of Islam mathematics, and which differ
from the classical Greek ones. For al-Khw arizm, as for most of his successors, any
kind of positive quantities arising from calculations, including fractions or irrational
roots, would count as legitimate numbers. In the last step of the argument, the side is
used to nd the value of the unknown magnitude. This approach clearly deviates from
Euclids consistent separation between geometric and arithmetic contexts, and it will
have signicant consequences over later developments. At the same time, however,
1 3
II.5 itself has not lost here its purely geometric character in any way. On the contrary,
al-Khw arizm is clearly implying that by reducing his problemto a geometrical context
he is bestowing theoretical legitimacy to his solving algorithm. He may have wanted
to appeal to certain readers with a more theoretical orientation, though perhaps not all
of his readers would think of this as a necessary requirement. It should be noticed that,
as in any algebraic solution of a quadratic equation, the crucial step in al-Khw arizms
procedure is that of completing squares, only that in this case, this completion is a
purely geometric procedure, rather than a symbolic, algebraic one as it will be much
later in the algebraic tradition of the seventeenth century.
An interesting question that has been a matter of lively debate among historians
concerns the sources of the ideas appearing in al-Khw arizms al-jabr wa-l-muq abala
and the degree of originality of his own contribution. This is true for both the algo-
rithms and the kind of geometric justications illustrated above. In the past, it was
common to assume Greek roots and a direct connection to the Elements, but more
recently, historians also started to indicate more prominently Indian and Central Asian
inuences. Following a different direction, Jens Hyrup has also suggested a possi-
ble connection with Babylonian traditions of problem solving that were alive and
inuential up until the European Renaissance (Hyrup 1986, 2001). One way or
another, ideas from Book II continued to appear repeatedly in later books of the
Arabic algebraic tradition, typically as part of a geometric justication similar to what
we have just seen with al-Khw arizm. Of the highest relevance to our discussion here
is the example of Ab u K amil (c. 850930) in his Kit ab f al-jabr wa al-muq abala,
9
a treatise written around 900. Ab u K amil presented in a systematic way methods
and results found in al-Khw arizm, while at the same time incorporating a visible
inuence of the arithmetic books of the Elements (Moyon 2007; Oaks 2001). His
treatise was widely read by European medieval mathematicians, and its inuence is
clearly visible, particularly concerning the questions that we are discussing in this
article.
Ab u K amil started by discussing the six cases of problems with squares as intro-
duced by al-Khw arizm. In providing geometric arguments to justify the validity of
his methods of solution, however, he followed the Euclidean source and its standards
much more closely than his predecessor. As a matter fact, in the text we nd for
most problems two geometric justications for each case: one closer in style to al-
Khw arizm and one relying directly on a result from Book II. This may reect a desire
to meet the requirements of two different kinds of readerships: one of practitioners
and another one of theoreticians. Still, in both cases Ab u K amil assigned numerical
values to lines and areas without any limitation, very much like al-Khw arizm had
done before him. Let us see the two proofs for the example of the square and ten
roots of the same square equal thirty nine numbers, where Ab u K amil relied on II.6.
This example, also taken directly from al-Khw arizm, is historically important since
9
A Latin version is extant which dates from the fourteenth century (Sesiano 1993). I will be referring to
this Latin text, which most likely reects what was available to the European mathematicians we shall be
discussing below. There is also a Hebrew version with comments by Mordechai Finzi (died 1475) which
seems to have been a translation from a Spanish version, but no such Spanish version has been preserved.
See (Levey 1966; Weinberg 1935).
1 3
658 L. Corry
it was repeated, with slight variations, by many mathematicians both in the Islam and
early European algebraic tradition (Dold-Samplonius 1987).
The accompanying diagram is the following (Sesiano 1993, 327328):
10
Figure 17
In the first proof, the square ABGD represents the square of the
unknown and to this the rectangle ABEU is attached and it is taken
to represent ten roots. This means that the line BH represents the
number ten since the rectangle share with square ABGD the line AB.
Now, in the argument rectangles and squares, as well as line segments
are taken to represent numbers, satisfying the conditions stipulated
in the problem, namely the square of the unknown, ABGD, together
with ten roots, ABEU, is thirty nine. Abu Kamils argument can be
schematically rendered as follows:
(e.1) R(AB,BE) +Sq(GB) =39 =R(EG,DG).
(e.2) But GD =GB, so that R(EG,GB) =39.
(e.3) Bisect BE at H. Accordingly, HB =5, and Sq(CB) =25.
(e.4) But BG is appended to HB in a straight line. Hence, [according
to II.6]:
11
R(EG,GB) +Sq(HB) =Sq(HG).
(e.5) Since R(EG,GB) =39 and Sq(HB) =25, then Sq(HG) =64.
(e.6) Thus, HG =8, and line HB =5. Finally, GB =3, and Sq(GB) =9.
The second argument relies on the equality of the rectangles RA and

ME, which by construction are both equal to rectangle AH. Hence, since
DE is 39, then the three surfaces, MB, BD, DY taken together are 39.
In addition, AC is 25, since HB is 5, from whence it follows that GC
is 64, and GH is 8, and, finally, GB is 3.
In the rst proof just described, the square HGCR and the segment YB are not even
mentioned. Indeed, they are not needed. These are, as in Euclids proofs, invisible
gures that are only implicitly referenced (i.e., HGCR is the square on SG). Thanks
to the use of II.6, one does not actually need to draw the entire diagram in order to
follow the proof. The diagram is drawn in full, however, because in the second proof
it is needed in order to follow the argument. Ab u K amil describes this second proof as
one that is meant to explain the problem so that it becomes apparent to your eyes.
Indeed, the fact that the three surfaces MB, BD, DY taken together are 39 can be
10
See also Oaks 2011, 255256.
11
sicut dixit Euclides in secundo tractatu libri sui.
1 3
seen directly on the diagram, based on the way in which it was constructed. The rst
proof, in contrast, is based on a proposition in Euclid which gives a rule that is applied
blindly, as it were. The rst is perhaps stronger in logical rigor, but it lacks the visual
transparency that characterizes the second proof.
12
There is a basic tension clearly reected in these proofs, one which actually under-
lies the entire treatise. This is the tension arising from the combination, within one
and the same text, of results and methodological approaches taken from the arithmetic
parts of the Elements (as well as from other, earlier arithmetic traditions) with results
and proofs originally meant to deal with continuous magnitudes, such as those of
Book II. Arabic numbers were of a more general kind than those handled in Euclids
arithmetic books, so that the latter could not fully account for the rules of calculation
for which Ab u K amil was attempting to provide a theoretical account. The results
of Book II helped Ab u K amil complete the picture, but in doing so, he subsequently
affected the way in which this collection of results was conceived, while broadening
the scope of its intended applications (Oaks 2011). (And see also [LC2].)
We have seen how in these two mainstream texts of early Islam mathematics,
al-Khw arizms and Ab u K amils, an intrinsic connection between the emerging tech-
niques of Islam algebra and the theorems of Book II was made explicit. As part of this
fundamental connection, a clear conceptual hierarchy was implicitly reected whereby
geometry appears as a main source of mathematical certainty, whereas the newly devel-
oping proto-algebraic ideas receive full legitimation fromgeometry. Within this entire
picture, and in spite of the broader scope of ideas within which it came to be applied,
Book II appeared as an important source of mathematical reliability which did not
thereby lose its essentially geometric character. It is also in this way that it would
be eventually perceived by the many readers of these treatises in Europe medieval
mathematical culture.
3.3 Th abit ibn Qurra
Like al-Khw arizm, also Th abit ibn Qurra worked in Baghdad under the patronage of
the caliph. He was active in the second half of the ninth century, at a time when many
Greek texts started to be translated and incorporated into the available body of mathe-
matics known in the Arabic culture. Ibn Qurra, who as already indicated had edited one
of the early translations of the Elements, also discussed the solution of problems with
squares of unknowns quantities in a treatise entitled Qawl f tas
.
h
.
h
.
mas a il al-jabr
bil-bar ahn al-handasya (Account of the Correctness of Problems of Algebra by
Geometric Proofs), which is of particular interest for our survey here. This treatise
presented general solutions for three of the six standard cases previously treated by
al-Khw arizm (though without mentioning al-Khw arizm), and, like al-Khw arizms
text, it also provided geometric justications for each of the procedures presented. Like
Ab u K amil somewhat later, Ibn Qurra referred explicitly to the propositions of Book
12
Besides this method of solution for nding the thing, whose validity is proved in two different ways,
Ab u K amil also introduced a second method of solution directly yielding the value of the square. Although
highly interesting in itself, it is beyond the scope of the present article. See (Sesiano 1993, 329330; Moyon
2007, 310).
1 3
660 L. Corry
II of the Elements as his basis for legitimation. But, unlike these two mathematicians,
Ibn Qurra discussed various cases in an abstract manner that did not involve specic
numerical instances (either in the formulation or in the solution). All the while, he
translated each step of the procedures he discussed into a corresponding component
of the geometric diagrams of II.5 and II.6. In this way he remained at the abstract level
that involved reference to general squares, roots and numbers, rather than to speci-
cally measured ones. This can be better understood by looking at the case he veried
with the help of II.5, which is the already mentioned case: square and numbers equal
roots of the same square.
In his presentation Thabit referred to the following diagram (Luckey
1941, 106107):
Figure 18
The square of the unknown quantity is represented here by the square
ABDG, while the line AH is measured by as many units as there are
roots in the given problem. By completing the rectangle GH we obtain
a figure that represents the number of roots, and if we subtract the
square BG, then DH represents the numbers in the problem. From here
the argument proceeds as follows:
(f.1) DH equals AB times BH, whereas AH is also known. In other
words, what we have is a given line AH, which is cut at B, in
such a way that R(AB,BH) is known.
(f.2) Bisect now AH at W. By II.5, R(AB,BH) +Sq(BW) =Sq(AW).
(f.3) But AW is known (since AH is known) and hence Sq(AW) is known,
and also R(AB,BH).
(f.4) Thus, we also know now Sq(BW), and hence we know BW.
(f.5) If BW is subtracted from AW (as in the figure on the right)
or added to AW (as in the left), then we obtain AB as a known
quantity, which represents indeed the root we are looking for.
Th abits algebraic procedure for solving this case is based, as we see, on complet-
ing a square and then adding or subtracting from the value of the number of roots
in the problem. This is quite similar to what we saw in the two previous examples.
When explaining the rationale behind the procedure in terms of geometrically com-
pleting the square on BW, the main trick comes from II.5. But Th abit did not just use
geometric methods in order to solve a specic problem, or to legitimate known proce-
dures. Rather, he essentially identied the two methods with one another by stressing
that his arguments are in accordance with those developed by the algebraists
1 3
(Luckey 1941, 95). This is a subtle step beyond the views implicit in al-Khw arizms
and later in Ab u K amils approach, and Th abits highly inuential status within Islamic
mathematics no doubt helped disseminate it. It thus seems reasonable to assume that
it may have contributed to strengthening views of Book II as a collection of tech-
niques directly related to the gradually developing algebraic procedures, rather than
just supporting it from outside.
3.4 Al-Nayrz
Another version of II.5 in the mathematical culture of Islam that is worthy of atten-
tion here is the one appearing in al-Nayrzs commentary to the Elements, dating
from the early tenth century, which was one of the earliest such commentaries to be
written in Arabic.
13
In his text, al-Nayrz incorporated idiosyncratic additions and
changes to the original text. Some of these comprised either a commentary on the
Euclidean proof or an alternative proof. Al-Nayrzs attributed them to Herons Com-
mentary. The most noticeable kind of additions to Book II were numerical examples
that illustrated each proposition. In the case of II.5, for instance, the numerical exam-
ple is of a line AB representing the number 10, which is then divided into two equal
parts, 5 and 5, and also into two unequal parts, 7 and 3. Thus, 5 5 = 25 equals
7 3 +2 2.
14
Asecond important kind of addition that al-Nayrz incorporated into the Euclidean
text appears inBookIX, where, as a series of comments toIX.16, he includedarithmetic
versions of I.1II.4 (Curtze 1899, 204207). It is evident that merely by placing these
propositions in the context of one of the arithmetical books of the Elements, and
by reformulating these four propositions in terms of numbers rather than in terms of
lines, al-Nayrz was making a strong statement about the way in which the knowledge
presented in Book II, particularly concerning distributivity, should (or at least could)
be seen.
15
But al-Nayrz went much further than that, and he gave fully arithmetic
proofs of these four propositions, all of which embody properties of distributivity. He
proved them simply by counting and comparing the numbers of units in each of the
products involved (see [LC2] for details).
One mayspeculate, of course, about the possibilitythat all or some of these additions
and modications were suggested to al-Nayrz directly by his own reading of Herons
13
This text in its various extant manuscripts and translations has had a somewhat convoluted history
(Brentjes 2001b), which I will not delve into here. I will refer to the text as rediscovered and published in
Latin translation in (Curtze 1899). For details, see also (Lo Bello 2003a, 2009); (Tummers 1994).
14
In Curtzes late-nineteenth-century edition of this text, Al-Nayrzs commentary was accompanied by
a footnote in which he formulated the original Euclidean text together with an algebraic rendering of
the proposition (Curtze 1899, 94). Even here this algebraic rendering is anachronistic. Reinterpreting a
geometric result in arithmetic terms (as Al-Nayrz did) is not yet the same as reinterpreting it algebraically.
15
Neither Curtze norto the best of my knowledgeany other later commentator has called attention to
the remarkable fact that al-Nayrzs IX.16 is an addition to, and indeed a generalization of, Euclids IX.15,
and that, incidentally, an interesting feature of Euclids proof of IX.15 is its reliance on what seem to be
arithmetic versions of II.3 and II.4 (Heath 1956 [1908]), Vol. 2, 404405). Campanus, following al-Nayrz,
included arithmetic versions of propositions of Book II as additions to the same IX.16. I discuss this point
in greater detail in [LC2].
1 3
662 L. Corry
operational rendering of the Euclidean proofs in Book II. Herons approach was, as
already stressed, operational and yet fully geometric. But it lent itself quite easily to
a direct translation into arithmetic, in a way that Euclids proof did not. Al-Nayrz
may have as well been inuenced by some other, unmentioned texts that circulated in
late antiquity or in the Islam context. Thus, for instance, in an encyclopedic Arabic
text published around 960 by the Ikhw an al-S
.
af a fraternity, the Ras ail (or epistles),
we nd the same kinds of numerical examples (Goldstein 1965, 154157). The text
in this work betrays the clear inuence of Nicomachus Introduction to Arithmetic,
written in the rst century AD, and it also formulates the propositions of Book II
directly as propositions about numbers, rather than lines. However, since there are
no proofs at all in this text, we cannot say in more precise terms what the writer of
the text had in mind concerning the nature of numbers and their relation to geometry.
This text, moreover, was probably unknown to the medieval mathematicians discussed
below.
The important point, however, whichcanbe assertedwithcertainty, is that al-Nayrz
reached few, but inuential medieval readers, and in important cases his examples and
additions were cited in texts written by those readers. Indeed, al-Nayrzs text was
translated into Latin by Gerard of Cremona (c. 11141187), who used the Latinized
version of the name, Anaritius (Lo Bello 2003a).
16
This Latin version does not seem
to have been widely read in medieval Latin Europe, but we know that it was seriously
studied by at least two important scholars: Roger Bacon (c. 12141294) and Albertus
Magnus (c. 12061280) (Hogendijk 2005). Inuenced by the text, Albertus wrote his
own commentary of the Elements, comprising a paraphrase of Books IIV as well as
philosophical discussions (Lo Bello 2003b; Tummers 1980). These two authors, who
played an important role in the spread of Greek and Arabic science in Western Europe,
were certainly enthusiastic about mathematics, and Gerards translation of al-Nayrz
was an important source for their knowledge of geometry. But this knowledge of
geometrywas somewhat limited, andit canbe saidthat bothof themfailedtoappreciate
the importance of al-Nayrzs text. And yet, al-Nayrzs original text became indeed
inuential, at least via Jordanus Nemorarius and Campanus de Novara, as we shall
see below.
These are a very fewexamples fromamong a much broader mathematical literature
of the culture of Islam. They make it clear, however, that the ways in which the ideas
of Book II were handled at the time created and helped disseminate an image of
Book II in which the original, purely geometric character intended by Euclid had
already been already broadened. This is not to say that the propositions of Book
II lost their clearly geometric anchoring. Yet as they began to be incorporated in
texts that suggested, either implicitly or implicitly, that they are about properties of
numbers, at least to the same extent that they are of lines, the way was opened to
additional, far-reaching changes, including possible algebraic interpretations. One
may imagine that mathematicians of later times, say in the Renaissance, who learnt
their Euclid from texts inuenced by Islamic mathematics while being also trained
in algebraic methods, might naturally tend to read the propositions of Book II as
16
One interesting point of his Latin version is that in Al-Nayrzs additions to IX.16, products of numbers
are called areas (superciales).
1 3
embodying essentially algebraic ideas. In the mathematical culture of the Latin Middle
Ages, we still nd a more uid and less consolidated mixture of ideas of various kinds
around Book II. An analysis of the ways in which its propositions were understood is
very challenging and rewarding. The remainder of this article is devoted to such an
analysis.
4 Book II in the early Latin medieval translations of the Elements
Euclids Elements enjoyed a prominent position in the intellectual world of the Latin
West. Withinthe historyof the transmissionof the Greekcorpus, noworkbe it math-
ematical, astronomical, philosophical, or medical, no matterhad anywhere near the
[same] number of Medieval versions (Murdoch 2003, 370). Versions, to be sure,
is the right term to be used in this context, since a large number of variants of the
Elements circulated in Europe that deviated in various degrees from what was con-
sidered the faithful translation. Moreover, many other contemporary mathematical
treatises incorporated portions, sometimes even relatively lengthy portions, or results,
of the Euclidean text. Thus, the story of the reception of the Elements in the Latin
West is a very complex one, which has been thoroughly studied by scholars and which
continues to offer a fruitful eld of historical research to this day.
It is generally accepted that prior to the twelfth century, only fragments of Euclids
text circulated in Latin Europe. The best known of these were attributed to a transla-
tion from the Greek by the late Roman philosopher Boethius (ca. 480524). They
circulated in relatively wide European circles, beginning in the ninth century, as
part of a compilation usually known in the scholarship as Geometria I, and then
in the eleventh century as part of a second compilation known as Geometria II.
Only fragments of the original Boethian translation have been preserved, but it is
known beyond doubt that it contained at least parts of Books IV. The translations,
at any rate, were far from unproblematic, and the texts did not contain proofs. Most
likely, the signicance of Euclids contribution was not really appreciated at this time
(Folkerts 1981).
During the incipient intellectual revival associated with the Carolingian empire and
in the centuries that followed, Boethius treatises and his idea of the Latin quadrivium
played a prominent role. The mediation of gures such as Cassiodorus (ca. 485ca.
585) and Isidore de Seville (ca. 560636) was of crucial importance as well. What is
of special interest for us here is that within the quadrivial image of the mathematical
sciences, it was arithmetic, rather than geometry, that had absolute primacy in terms of
disciplinary importance (Guillaumin 2012). In Geometry II, for instance, Boethius
described the knowledge of numbers as the high point of all philosophy (Folkerts
1970, 138). Cassiodorus, in turn, sawarithmetic as the origin and source of the other
mathematical sciences, which depend for their being and existence on arithmetic
(Eastwood 2007, 192). This theoretical primacy, moreover, was further enhanced by
the contemporary incipient interest in the practical aspects of arithmetic in day-to-day
life.
In the early twelfth century, an unprecedented amount of Greek treatises gradu-
ally became available in Latin, many of them via Arabic translations. Their impact
1 3
664 L. Corry
brought about signicant changes in the world of learning at large, in its contents
and in its institutions. This is the time when existing cathedral schools developed into
universities at Bologna, Paris, and Oxford, and when new universities were created at
Cambridge and Padua. While perhaps their overall centrality receded, the quadrivial
arts continued to be taught at these new universities, side by side with the new sub-
jects and texts introduced into the curriculum (Kibre 1981). The disciplinary primacy
accorded to arithmetic over geometry within the mathematical sciences, to be sure,
remained unchanged.
Translations to Latin from Arabic or Hebrew original works aroused no less enthu-
siasmthan those of Greek texts. But whenever their texts were available, Greek authors
were invariably seen as the main authorities and were considered to be more reliable
than their Islamic counterparts. The latter were more often than not used as support
or commentary with respect to questions of correct or false attribution to works of
the ancient Greeks. In the words of Hyrup (1998, 320): Arabic texts, even when
translations from the Greek, were treated as objects to be used, and not always very
formally. Greek texts, on the other hand, were sacred objects and handled as such;
they were normally translated de verbo ad verbum (so closely indeed that both the
number of words and their order were often preserved ).
Among the earliest translations of the Elements from Arabic, historians usually
count those of Hermann of Carinthia (c. 1100c. 1160) and of Gerard of Cremona
(already mentioned above in relation to al-Nayrz). These translations, however, were
not frequently read and copied. According to the currently accepted scholarship,
17
the
texts that provided the basis for the main corpus of manuscripts that spread throughout
Western Europe over the twelfth and thirteenth century were produced by Adelard of
Bath (. 11161142), by Robert of Chester (. c. 1150), and, somewhat later, by
Campanus de Novara (12201296). Roberts text is not a direct translation from the
Arabic but rather a compilation of different texts, among them the translations of
Hermann and Adelard. Campanuss text was the one that in 1482 appeared as the rst
printed version of the Elements and which became the standard source of reference
for the Latin world until the sixteenth century when new versions were printed, based
on direct translations from the Greek.
18
The issues that have attracted the attention of historians investigating the Euclidean
tradition in the Latin Middle Ages must be understood against this background. A
main such issue concerns the ways in which the Adelard-Robert versions and the
Campanus version of the Elements (and their derivatives) differ fromEuclids original.
Thus, unlike in Euclids original, one often nds in these texts a careful labeling of the
various sections of the proof (exemplum, dispositio, ratiocinatio, conclusio, exempli
gratia, rationis causa, etc.), direct references to earlier propositions on which a proof
is based, and hints about other propositions which are directly related to the one
17
The introduction to (Busard 2005) contains a clear and well-organized summary of the rich and scholarly
impressive literature that has dealt over the last thirty years with the issue of the Latin versions of the
Elements. The reader may also consult the following: (Brentjes 1997/1998, 2001a); (Busard 1983, 1984,
1996a); (Busard and Folkerts 1992); (Folkerts 2006Section III); (Murdoch 1968, 1971).
18
Murdoch (1966) called attention to a very accurate twelfth-century translation of the Elements by an
anonymous translator who also translated Ptolemys Almagest fromthe Greek. However, there is no evidence
of any substantial use of this translation. See also (Busard 1987).
1 3
considered. On the other hand, rather than explicit proofs many texts provide only
hints to the main ideas on which the arguments are based. Differences in wording and
in the level of detail created the impression, shared by many at the time, that Euclid
was author only of the denitions, postulates, axioms and enunciations, whereas the
proofs were considered to be just commentaries provided by translators or compilers
(Busard and Folkerts 1992, 16).
Murdoch (1971, 2003, pp. 372 ff.) has stressed three main concerns that were
prevalent in the medieval versions of the Elements: didactic considerations, clari-
cation of the logical structure of the entire treatise, and the relation of the math-
ematical contents to more general philosophical questions and conceptions. Two
topics that received special attention in this regard were incommensurable magni-
tudes and the nature of the continuum (Murdoch 1968, 9093). Accordingly, the
existing scholarship has dealt mainly with the ways in which the Latin medieval
tradition handled the theory of ratios and proportions, and particularly in relation with
Book X (see, e.g., Rommevaux 2008). The relevance of these topics to Book II and
to the questions discussed in this article, in turn, has typically received much less
attention.
Of particular importance for our discussion here is the fact that the rigid separation
between the realms of number and of continuous magnitudes consistently preserved
in the Greek original was highly unsatisfactory for medieval mathematicians. This is
interestingly manifest throughout the texts of the various Euclidean versions. Thus,
for instance, while in his arithmetical books, VIIIX, Euclid strictly refrained from
using the Eudoxian theory of proportions of Book V, we nd instances of its use in
the Latin versions, together with interesting comments related to them. We also nd
proofs of propositions about continuous magnitudes based on the use of particular
numerical examples, for instance in Books V and X. We have already seen a similar
use of numerical examples in the text of al-Nayrz, but in the medieval versions,
we also nd arithmetic ideas used in the proof. Moreover, in some Latin versions of
the Elements, the only argument given was that of a numerical example. Likewise,
we can also nd examples of geometric arguments and concepts being used when
discussing some propositions in the arithmetical books (for details, see Murdoch 1968,
8690).
And yet, in spite of the importance accorded to arithmetic in the medieval mathemat-
ical landscape, and the other kinds of differences with the Greek original mentioned
above, most proofs of Book II found in the Latin versions remained very close to the
original Euclidean ones, or at least to their Heronian variants. It is fair to say that in
general, no arithmetic or algebraic ideas were introduced at the core of the proofs of
Book II and that the purely geometric character of the propositions was essentially
preserved (the Campanus version presents a very interesting turn in this regard, but I
discuss his case separately).
We can illustrate the situation by looking at how II.5 appears in the various texts
in the Adelard-Robert tradition. First we can look at the so-called Adelard I ver-
sion, which is most likely the rst complete translation from an Arabic version of
the Elements. We nd there exactly the same geometric proof as in Euclid, based
on the use of the gnomon (Busard 1983, 7576). Also the proof of II.5 appearing in
the so-called Adelard II versionnowadays considered to be a compilation attributed
1 3
666 L. Corry
to Robert of Chesterremains close to the purely geometric spirit of the Greek origi-
nal, albeit without repeating all the details found in Euclid. Many extant manuscripts
of this version exist, and its enunciations were used by many other Latin commen-
tators or editors, including Campanus. Many of the proofs are typical examples of
providing no more than general indications about the argument, and in some of the
manuscripts there are only diagrams. For II.5, the guidelines refer to the following
gure and stipulate the following steps (Busard and Folkerts 1992, 132):
19
Figure 19
Asquare is constructed on the middle of the whole line. Through the point of the
unequal section draw a line segment parallel to the side of the square, and then a
diagonal in the square. Where these two lines intersect drawanother parallel line
[and use it to complete a rectangle]. Two equal complements are obtained, and
[the rectangle] contained by the two unequal portions equals the whole gnomon.
Hence, the residue between the two gures is manifestly the square [built on the
segment between the two cuts].
The general indications for proofs appearing in the Robert version were subse-
quently elaborated in various ways. Thus, for instance, in a thirteenth-century adap-
tation of the text, the proofs of Book II preserved their essentially geometric spirit,
but were elaborated along the cut-and-paste approach introduced in Herons proofs
[see the proof of II.5 in (Busard 1996a, 102103)]. The geometric proof based on
the gnomon also appears in the translations of Hermann of Carinthia (Busard 1968,
4445), Gerard of Cremona (Busard 1984 4344), and in the translation made directly
from the Greek (Busard 1987, 5859).
And yet, in spite of this, some alternative to the proofs of Book II, and specically,
some arithmetically oriented proofs did circulate in the Medieval Latin world. We nd
them in books other than the Elements, or, interestingly, in the case of Campanus,
as part of additions and comments to the three arithmetic books of the Elements.
The next (and last) section of the article is devoted to these ideas, as they appear in
the works of medieval scholars such as Abraham Bar H
.
iyya, Leonardo Fibonacci,
Jordanus Nemorarius, Campanus, Gersonides and Barlaam.
19
Unless otherwise stated, translations from Latin, Hebrew, or German are mine.
1 3
5 Book II in other medieval texts
5.1 Abraham Bar-H
.
iyya
I start this survey with Abraham bar H
.
iyya ha-Nasi (ca. 10651145), also known
as Savasorda. He was a Jewish sage born in Barcelona, and probably educated in
Zaragoza. Little is known of his life, but the importance of his work for the transmission
of Arabic science to the Jews in Spain and to the Christian Latin West in general
is widely acknowledged. He wrote Hebrew treatises in various elds of knowledge,
such as philosophy, astronomy and astrology. His treatise, H
.
ibb ur ha-meshh
.
ah we-ha-
tishboret ( ), a title often translated as Treatise on Measurement
and Calculation, was a book on practical geometry (Lvy 2001, 3742; Sarfatti 1968,
64128). A Latin version by Plato of Tivoli appeared in 1145 under the title of Liber
Embadorum. This Latin text introduced for the rst time in Europe the techniques of
Islamic algebra for solving quadratic equations, thus antedating Gerard de Cremonas
Latin translation of al-Khw arizms Algebra. The Liber Embadorum was widely read
and it is known to have directly inuenced Leonardo Fibonacci (Clagett 1978, 1265;
Curtze 1902, 57; Mills-Vallicrosa 1931).
Bar H
.
iyya is the earliest representative of a tradition involving Jewish sages who
assimilated important parts of the Greco-Islamic scientic culture and published their
own works in Hebrew, mainly between the twelfth and fourteenth centuries. Such
works were intended as a vehicle for transmitting Arabic science to Jewish audiences
not conversant with the Arabic language. They often comprised compilations and
systematic presentations of previous work, but in many cases they also presented
original ideas (Lvy 1997a,b,c). Characteristic of many of these texts is the noticeable
effort made by the authors to stress their nature as practical treatises that draw their
contents from the wisdom of the nations ( ) but which, at the same time,
provide useful tools in the service of the teachings of the Jewish law. In the introduction
to Bar H
.
iyyas H
.
ibb ur, for one, he explicitly stressed as a main motivation, what he
saw as the existing incompetence typical of many sages of his generation when it
came to implementing the Jewish laws of inheritance that involved calculations of
areas and division of patrimonies (Guttmann 19121913, 34). One should keep in
mind, however, that in many Hebrew texts of the time we nd this kind of formulation
used as a rhetorical device to justify the fact that a learned Jewis devoting his precious
time and efforts away from the study of the Bible.
As Tony Lvy has stressed, we do not currently posses an accepted critical version
of the Hebrew text of the H
.
ibb ur. Nor has the issue of Bar-H
.
iyyas sources been duly
claried. Still, it is remarkable that neither the termalgebra nor the standard technical
vocabulary normally used in the Islamic algebraic tradition appears in his text. On the
other hand, the text does include much of the techniques for solving problems involv-
ing squares of the unknown and the kind of geometric justication that was standard
since the time of al-Khw arizm. Thus, it seems reasonable to assume that Bar-H
.
iyya
knew well the relevant literature, but that he wanted to keep the technical vocabulary
to a minimum so as to make his text easier for his intended readership to comprehend
(Lvy 2001, 5152). For the purposes of the present article, it seems sufcient to put
aside the textual difculties that future scholarly work will have to further elucidate
1 3
668 L. Corry
(Lvy 2001, 4547), and to refer, without further comments, to the Hebrew text pub-
lished in 19121913 in Berlin by Guttmann (18721942).
20
Bar-H
.
iyyas book has four main parts. In the rst one, we nd a collection of
geometric denitions and some elementary propositions relating to them. These are
intended as providing the tools for what will be done in the following parts of the
book. Most of the results presented in this section paraphrase results drawn from
the Elements, including Book II, as we shall see right below. In the second part of
the H
.
ibb ur, Bar H
.
iyya discussedquestions relatedtomeasurements of lands, buildings,
and elds having all kinds of shapes: quadrilaterals, triangles, circles, and some more
complicated ones. This is, indeed, the most important part of the book. The third part
deals with the division, into elementary gures, of those considered in the previous
section. Finally, in the fourth part, Bar H
.
iyya extended some of the previously obtained
results into volumes: prisms, cylinders, spheres, etc.
Given the main focus of interest of the book as developed in its second part, it
is not at all surprising that Bar H
.
iyya mixed in a very conscious and purposeful
way ideas taken from both geometry ( ) and arithmetic ( ). Thus,
while the rst twenty denitions introduce basic geometric concepts, going from the
point to the line (both of them following the Euclidean denition), and to the areas
of various types of polygons, the following ones dene the basic arithmetic concepts.
The latter essentially repeat those of Euclid in the introduction to Book VII. This
is the case, for instance, with BH-21,
21
where unit and number are dened: The
unit is that thing whereby each thing in the world is called one. And number is a
collection of a multitude of units. But in some other cases, we also nd minor, thought
interesting differences that stress Bar H
.
iyyas basic intention to apply these concepts
in a framework that mixes the geometric with the arithmetic. Thus, while Euclid had
used term such as plane numbers, square numbers, or cube numbers, in a rather
metaphorical sense and while he never represented such numbers in his propositions
other than as straight lines, Bar H
.
iyya, to the contrary, added to the corresponding
denitions, a small diagram in the spirit of the Pythagorean gurate numbers. In this
way, he prepared the ground for organically incorporating arithmetic into geometry.
Thus, for instance, in the following two examples:
BH-22: And the number that is multiplied by another number is the number that
is added to itself as many times as there are units in the second number in which it
is multiplied, like two times three or two times ten and it is called a plane number
and this is its shape ::: and the number obtained by this addition is called a plane
number.
BH-23: And a square number is a number obtained by adding a number to itself
as many times as there are units in it, and the rst number is called the root of
the square ( ). For example the number nine is called a square number
20
The Hebrew printed version of Bar H
.
iyyas text is accompanied by an introduction in Hebrew written
by Guttmann, and it contains further comments, also in Hebrew, written by Zvi Hirsch Yaffe (18531927).
The Guttmann edition also served as the basis for a translation of the text to Cataln (Mills-Vallicrosa
1931).
21
The paragraphs are numerated in the edition I am using here. I add the initials BH, for ease of reference.
1 3
because it is obtained from adding three to itself three times, and the number three
is called the root of the square and this is its shape
.
.
.
.
.
.
.
.
. .
22
Bar H
.
iyya also dened a cubic number ( ), but not merely in a gurative
manner, as one whose units can be arranged as a cube. Rather he dened it as a body
( ) of equal length, breadth, and height.
After this arithmetic preliminaries, Bar H
.
iyya moved without any clear separation,
to work out several propositions which provide the main tools underlying the solutions
for the problems discussed in the second section and in which arithmetic and geometric
statements and arguments are freely mixed together. The connecting link is provided
by a general comment, to the effect that the square of a segment is that gure built when
another segment of the same length is set perpendicular to the given one and a square
is thus built with four equal sides, while a rectangle is similarly built on two different
segments. From here on, it is unproblematic to speak about geometric construction
while associating numbers to the lengths of the segments. And this is indeed what
Bar H
.
iyya did when presenting propositions II.4II.10 (but not including II.7), all of
which deal with ways of dividing a segment into two parts and constructing squares or
rectangles on the parts obtained (BH.27BH.32). It is worth examining in some detail
how he formulated and proved these propositions, while paying special attention to
the unique combination of arithmetic and geometric ideas that he adopted.
Consider, for instance, BH-27, which corresponds to Euclids II.4. Bar H
.
iyya speaks
here about a line that is divided at an arbitrary point and about the squares that are built
on the resulting segments. But to this purely geometric formulation he added: and I
give you an example with numbers ( ), which in this case
is a line of length 12 divided into two segments of length 7 and 5. We can look more
closely at the specic roles that Bar H
.
iyya assigned to geometry and to arithmetic
in his presentation, by examining in closer detail what he does in the proposition
corresponding to II.9 (BH-31). Let us rst look at the proof as it appears in the text
and then I will comment on it.
The proposition is formulated in the standard Euclidean way and it
is also exemplified by a line of length 12 which is divided at its
midpoint and at another point into segments of lengths 7 and 5.
23
Two squares taken together, one of side 7 and one of side 5, make
74. And on the other hand duplicating the square with side 6 and the
square with side 1 (1 being the difference between the half, 6, and
the long segment, 7) also makes 74. Having introduced this numerical
example, Bar H
.
iyya moved to the proof, which interestingly differs
from Euclids, whichas seen aboverelies in a non-trivial way on the
Pythagorean theorem. Bar H
.
iyyas own wording is quite contrived but
it is worth being presented here (with some simplification). Below I
will streamline his argument with the help of diagrams, which will
make it easier for the reader to capture the main idea.
22
Read from the point of view of current Hebrew usage, Bar H
.
iyyas terminology may give rise to some
confusion, as was pointed out in (Sarfatti 1968, 82). I am following here Sarfattis interpretation (e.g.,
multiplication for , addition for , root for ). For a brief account of the differences between
ancient and modern Hebrew mathematical terminology, see (Corry and Schappacher 2010, 449457).
23
Actually, the text in Guttmann (19121913), p. 15, says: (on 4 and on 5). But this seems to
be a typo.
1 3
670 L. Corry
Take the line AB, of length 12, and divide it in its half at E, and
also into two different segments, of length 7 and 5, at I.
Figure 20
Construct AICD being 7 on 7, and BIGF being 5 on 5. Taken together,
says Bar H
.
iyya, these two squares equal the sum of AEMK and EMJB
(which is the same as taking twice the square on the half, that is
twice 6 on 6), together with HNLC (which is twice the square of the
difference, that is 1 on 2). And why this is so? Becausetells us Bar
H
.
iyyaFGHJ is placed under BIGF and on the same line with the square
of half the segment, AEMK.
24
And we count it in this way together with
square AICD. Thus, both FGHJ and NKDL are like 5 on 1. And the reason
for this is that AKD equals AEI and both are 7 and on the other hand
AK equals AE which is 6, and hence what is left is KD which equals
IE which is 1. But also DLC equals AEI and if one subtracts CL which
is 2, then one is left with LD which is 5. Hence NKDL is 1 on 5 and
the same is true for FGHJ because MJ equals EIB which is 6, and the
line BF equals BI and we are left with FJ and IE which are 1, and the
length of JH equals that of BI. Here ends Bar H
.
iyyas proof.
Now, just from this more or less literal transcription of the proof, we can already
learn much about Bar H
.
iyyas overall approach. In the rst place, in spite of his insis-
tence in repeating the numerical values of the segments at various places along the
proof, these values play no role whatsoever in the argument and they are never arith-
metically manipulated in order to get any result. Rather, they only accompany the
steps of an argument that fully relies on completing a geometric construction and on
learning from looking at this construction. Indeed, the geometric construction is quite
simple and what is said about it does not go beyond basic cut and paste considera-
tions. But in addition, since Bar H
.
iyyas wording for this particular proposition is so
contrived (this is much less the case for other propositions in this section), one may
even speculate that without introducing here the numerical values of the lengths Bar
24
(Guttmann 19121913, 15) [In the quotation, I have replaced the Hebrew letters
appearing in the original diagram with the letters I am using here in my diagram]:
1 3
H
.
iyya may have had some difculty in working out the argument even for himself.
More certainly, he may have thought that his readers would nd the argument easier
to follow in this way, and in any case, in the later sections of the treatise, he would be
using the result only in its arithmetic version.
But it is also important to notice that, perhaps because of the didactical character
of this treatise, Bar H
.
iyya often relied on the diagrams in a stronger sense. That is,
in constructing the gures that appear in the diagrams of many of his proofs, he left
implicit several facts that he nevertheless used as part of the argument. The facts have
to be inferred from the diagram while following the steps of the proof. Thus, in the
case we are examining here, the diagram showsbut Bar H
.
iyya does not say sothat
AK = BJ = AE, and hence HC equals IE. Also, part of the construction, as we learn
only from the proof itself, is that HN equals twice IE.
Keeping these general remarks in mind, we can now reformulate the core of the
proof by looking at the diagrams and reasoning with them.
With reference to the diagram, what we are asked to prove is just:
AICD +BIGF = 2Sq(AE) +2Sq(IE) .
Bar H
.
iyya tells us that in his construction CL is 2, and this means
that HMN = 2IE, and hence HNLC = 2 Sq(IE).Thus, all what remains to
be shown is that NKDL = FGHJ. And this is true, he says, because the
long side of both rectangles equals BI, while their shorter sides
equal IE. Let us see why this is so, starting with the short sides,
KD and FJ, and stating the idea in purely geometric terms:
(g.1) By construction AKD =AEI and AK =AE, hence AKD-AK = AEI-AE,
hence KD =IE
(g.2) Again by construction JHM =EIB and BF =BI. But BI =FG =JH,
hence JHM-JH =EIB-BI, hence FJ =IE.
(g.3) Hence [from (i.1) and (i.2)]: KD =FJ.
And now I give the argument for the long sides, HJ and LD, using
numerical values as Bar H
.
iyyas did, because in the last step one
sees that, although there is nothing here but a simple geometric
situation that holds true independently of the values, explaining it
with the numerical values really makes it easier to follow:
(g.4) By construction HD =BI (and both are 5this is left implicit
using the fact that FGHJ is placed under BIGF).
(g.5) DLC =AEI and if one subtracts CL which is 2, then one is
left with LD which is 5.
CL is 2 means, of course, that CL was built as twice the difference
between the longer segment and the half of the given segment, even
though this was not explicitly stated. But explaining the situation
in these more abstract terms would make the argument longer, and
apparently Bar H
.
iyya was happy with being able to make it shorter and
easier to follow with the help of the numerical values.
Notice that the main tool on which the proof relies (and this is true for various other
proofs as well) is the distributivity of the product (or, more precisely, of rectangle
formation), which as we knoware embodied in II.1II.3. Bar H
.
iyya applied freely this
property without in any way discussing it or even mentioning in the text.
1 3
672 L. Corry
Bar H
.
iyya completed the rst part of his book by discussing (in BH-33 to BH-41)
various propositions taken mostly from the Elements,
25
and which deal with proper-
ties of circles, triangles, and other gures. And as already stated, part one is actually
intended as providing tools for solving numerical problems involving an unknown
quantity. Hence, I complete nowthis overviewof Bar H
.
iyyas text by showing an exam-
ple of solving a quadratic equation (in BH-49) while using II.5 for justifying the proce-
dure used. The problemis the following: if the square is subtracted fromfour times the
side the result is three ( ).
The solution is found by following a simple procedure: half the number
of sides is 2 whose square is 4. Subtract from this the remainder
three and you obtain one, whose square is one. Subtract this result
from half the sides and you are left with one, and this is the side of
the square, or add the result to half the sides and you get 3 which
is also the side of the square. The side may be one or it may be 3,
because there are two different calculations for this question.
26
As in the Islamic tradition, Bar H
.
iyya also presented geometric proofs for justi-
fying the procedure used in solving the problem. It is worth seeing here some of the
details of this proof that are highly reminiscent of Th abit ibn Qurra.
The proof is better understood by reference to the two figures
that accompany the text, of which the following is the first one
( ):
Figure 21
In this figure, AB is taken to be four amoth ( ), the coeffi-
cient of sides in the problem, whereas AC is the side of the
square referred to in the problem. The argument then goes as follows:
(h.1) Subtract from the parallelogram ACDB a square ACFE; the
remaining parallelogram EFDB is three amoth, [this is actu-
ally the statement of the problem]
(h.2) Cut now AB into two equal parts at G; since E divides AB into
two unequal parts, it follows that the parallelogram on AE,EB
together with the square on EG equals the square on AG, [this
is II.5]
25
These include, in the order in which they appear, versions of: III.35, I.33, VI.4, I.37, I.38, I.41, I.35,
I.36, and VI.1.
26
In Sect. 49: . In the text Bar H
.
iyya sometimes calls the number by its
name and sometimes uses the alphabetic characters to indicate them. For the side of the square
or for the unknown Bar H
.
iyya uses alternatively and even within the same sentence. The text
has many interesting linguistic aspects that I do not discuss here, starting with the very use of the word
thisboreth ( ).
1 3
(h.3) But we know that the square on AG is 4 [since G cuts AB in its
middle point], and if we subtract from it the parallelogram
on AE, EB (namely, EFDB) which is 3, then we are left with 1,
and this is the value of the square on EG,
(h.4) Hence EG is one, and, since AG is two, EA is one, and this is
the side that we are looking for.
As part of the same proof, Bar H
.
iyya also indicated how the second
solution is justified by reference to a second diagram( ):
Figure 22
The argument is similar to that of the first part, and we need not go
here into all the details.
This problem is similar to Ab u K amils treatment of square and numbers equal
roots of the same square. Also Ab u K amil justied the procedure for solving this
case with the help of II.5, and he accompanied his explanations with two diagrams
similar to those of Bar H
.
iyya here. Curiously, however, Bar H
.
iyya presented his
problemby subtracting the square, thus departing fromthe practice followed in Islamic
texts (to state it in modern algebraic terms: Ab u K amil used the typical formulation,
x
2
+b = ax, whereas Bar H
.
iyya formulated the problem as: ax x
2
= b). And, as
did Ab u K amil, Bar H
.
iyya indicated two solutions for the problem.
Bar H
.
iyyas explicit and implicit references to results originating in Book II provide
us with an illuminating, direct insight into one specic (and perhaps rather idiosyn-
cratic) way in which a reader with a signicant knowledge of Islamic techniques of
problemsolving could read Euclid froman arithmetic and proto-algebraic perspective.
Bar H
.
iyya understood the propositions of Book II as both an expression of properties
of numbers and as the proper way to provide legitimation for algebraic techniques
of problem solving. But he continued to read these propositions as ultimately based
on geometric reasoning, even when he thought it convenient, possibly on didactical
considerations, to accompany with numbers the text where he described this pure geo-
metrical reasoning. Products were seen here as square or rectangle formation, and the
proofs were based on geometric properties of these gures.
The readership of Bar H
.
iyyas Hebrew text, of course, was rather limited, and
it thus had little direct impact on the wider Latin medieval mathematical culture
in Europe. Nevertheless, as already indicated, the text also appeared in 1145 in a
Latin version commonly attributed to Plato of Tivoli which did have an important
impact, mainly through the work of Fibonacci. As Tony Lvy has indicated, one may
1 3
674 L. Corry
conjecture that in preparing the text, Plato worked with some Jewish collaborator,
possibly (but by no means certainly) Bar-H
.
iyya himself (Lvy 2001, 5355). The
Latin version, at any rate, is not just a mere transcription, and it differs from the
Hebrew text in various senses (omissions, additions, reformulations, etc.). Of inter-
est here are some differences pertaining to the interplay between geometry and
arithmetic.
27
Such differences appear, for instance, in the rst part of the text. The Latin version
contains no diagrams at all in this part, contrary to what is the case in the Hebrew
one, but on the other hand, it follows much more closely the structure and wording of
Euclids Book I. Indeed, here as in other parts of the text, results from the Books IIV
which are introduced without proof and which are necessary for proving other results,
are clearly not directly translated from Bar H
.
iyya. Tony Lvy (2001, 54; note 3) has
suggested that they were taken from an Arabic version of the Elements, possibly of
the more ancient ones associated with Hajjaj.
As an example of the difference between Plato and the original text of Bar H
.
iyya,
we can notice that the latter introduced parallel lines ( ) quite early
in the section and that he gave in the same paragraph two different denitions, namely
as equidistant lines (i.e., as such lines that the distance between them is constant
also when they are extended) and as lines that do not intersect if extended innitely
( ). In the Latin version, in contrast, the denition appears at the end of the
preliminary section (Curtze 1902, 15, BH-28), and exactly as in Euclid: These are
lines that, while laying on the same plane, do not meet with each other when produced
innitely (ad innitum).
After the preliminary denitions, we nd in the Latin version the ve Euclidean
postulates and a list of common notions. None of this appears in the Hebrew text.
The common notions are not just the ve Euclidean ones, but also some additional
ones, such as two things which are the double of the same thing are equal to each
other, and the same for the half, or a thing is the sum of its parts and two lines
contain no area (Curtze 1902, 1517). Another important difference concerns the
basic denitions of numbers and their properties, which in the Latin version basically
repeat those appearing at the beginning of Book VII. At the same time, denitions such
as BH-21, BH-22 of the Hebrew version (see above) are absent from the Latin text.
In the Latin version, the propositions of Book II are only formulated and not proved,
since they were already proved by Euclid (p. 19). Still, the geometric formulation
of each proposition is followed by a numerical example. On the other hand, it is
interesting to notice that in the second part of the treatise, the problems solved with
the help of the propositions of Book II are essentially a verbatim translation of the
Hebrew original.
One canonlyspeculate about the reasons for differences betweenthese twoversions,
and to attribute them to the differences audiences and different purposes against the
background of which they were conceived. One may also hope for new, critical edition
of Bar H
.
iyyas text on the question of the complex relationships between the Hebrew
and the Latin texts.
27
Like with the Hebrewtext, also here there are some open issues concerning the manuscripts. See (Curtze
1902, 4).
1 3
5.2 Liber Mahameleth
Liber Mahameleth is a text on commercial arithmetic, presumably written in or near
Toledo around 11431153. Anne-Marie Vlasschaert, who recently published a critical
edition (Vlasschaert 2012), has suggested that Dominicus Gundisalvi (c. 11101181),
or someone in his school of translators, may have been the author. The text bears
some relation to an earlier Arabic text likely called Kitab al-muamalat (Book of
Transactions). Most likely, however, Liber Mahameleth is not a direct translation
from the Arabic, but rather a new work written in Latin in the context of the scholarly
culture of Al-Andalus.
The author evidently had a good command of Euclids Elements and of Abu
K amils Algebra, and he constantly relied on their results within his own proofs.
Liber Mahameleth can be seen as an attempt to provide a comprehensive synthesis
of important contributions of these authors, as well as of Nicomachus, al-Khw arizm,
and some other classical Greek and Arabic authors. This synthesis, moreover, was
specically oriented toward providing a practical compendium of arithmetical tools,
grounded on a systematically discussed theoretical basis for the basic operations, par-
ticularly those involving fractions. The core of the text deals with applications of these
operations to problems related to commercial transactions, such as prices, prot shar-
ing, consumption of oil by lamps, of feed by animals or of bread by humans, etc., as
well as of problems of a more theoretical kind.
The rst ten propositions of Book II of the Elements appear prominently in the pre-
liminary, theoretical section of Liber Mahameleth. Clearly inuenced by Ab u K amil,
the author certainly went much further than him in positioning these propositions in
a foundational role for arithmetic, and in explicitly stressing their indispensability for
any prospective reader of the book. Indeed, the author stated that he was reformulating
what in Euclid are propositions on segments, and that he was rephrasing them now
for numbers (p. 25). Still, on closer examination and in spite of this stated intention,
most of the proofs provided for propositions taken from Book II retain a distinct geo-
metric avor. This can be seen, for instance, in relation with the proof for II.5 (p.28),
which essentially repeats Herons cut-and-paste argument. As we saw, Heron operated
upon areas of rectangles and squares (i.e., rst separating them into components and
then reassembling), and also here, in Liber Mahameleth, the argument explicitly relies
on applying distributivity of rectangle formation over addition, such as embodied in
the corresponding versions for II.2II.3 (pp. 2627). Also the accompanying diagram
is similar to Herons:
Figure 23
Again as with Heron, we do not nd here actual geometric constructions made on
the segments, but squares and rectangles are indeed implicitly constructed as part of
the reasoning. In all respects, the proof qualies as geometric in the sense that I
have been using here, and in none of the steps there is anything like counting units
comprised by any of the numbers involved.
1 3
676 L. Corry
What is of special interest in Liber Mahameleth in relation with Book II is that in
laying down the foundational aspects of arithmetic, the author combined this kind of
propositions and proofs with others that are purely arithmetic, thus giving rise to an
original blend of approaches, in which distributivity of the product was assigned a
pivotal role. Proposition 9 of the preliminary section (pp. 2526), which is parallel to
Euclids II.1, plays an important role in the presentation. Its proof is worthy of special
attention, because it is based neither on geometric nor on arithmetic considerations,
but rather on properties of proportions drawn from Books V and VII. I discuss this in
greater detail in [LC2].
It is also pertinent to see how the propositions appearing in the preliminary section
of Liber Mahameleth are then used to justify procedures for problem solving in the
main body of the book. Thus, for instance in the following example which uses a
version of II.6 (pp. 410411):
A ladder of unknown length is leaning on a wall, and it is withdrawn from its
bottom by a distance that, when the descent from the top is subtracted from it,
the remainder is 4, and if we multiply those two lengths the result is 12. What is
the length of the ladder?
28
The situation is represented in the following diagram:
Figure 24
The proof starts with the data of the problem, namely ad bg = 12 and
bgad=4. Now, the length hg is cut from bg, so that hg=ad. Hence,
bh=4. But also bghg=12 (since hg=ad). Finally, the point z bisects
bh (i.e., hz = 2). Here, we can apply II.6 (sicut euclides dixit)
and we obtain: bghg+hz
2
=zg
2
. Thus, clearly, zg=4, and consequently
also bg=6. Finally, ad=2.
This is certainly not an example of an application of the propositions of Book II to a
real-life problem, but what really concerns our discussion here is how the author of
Liber Mahameleth, also in this more applied part of the text, combined the arithmetic
and geometric ideas discussed in the preliminary section. Very much like the didac-
tical approaches followed in the contemporary versions of the Elements we nd here
continual cross-references between the two realms without the author emphasizing the
possible differences. This creates a somewhat ambiguous but very interesting situation.
28
A detailed analysis of this problem and its roots in Babylonian mathematics appears in (Sesiano 1987).
1 3
Indeed, notice that in my rendering of the proof above I wrote the products as arith-
metic, not geometric operations. Thus, for instance, I used II.6 as bg hg +hz
2
= zg
2
,
rather than as I did with all previous authors, R(bg, hg) + Sq(hz) = Sq(zg). There
were actually good reasons to choose the geometric rendering. For one thing, the entire
situation is embodied in a diagram. For another thing, the wording for the product here
is exactly the same as that used for rectangle formation in the rst part of the treatise
(Quod igitur t ex ductu bg in hg et hz in se equum est ei quod t ex ductu zg in se).
Indeed, it is precisely this same wording which allows the use of the theoretical results
proved in the rst part for handling specic problems appearing in the second. Never-
theless, the choice of arithmetic rendering was guided by what seems to have been the
authors own view, namely that he was dealing in this part with numbers. But the choice
could have been in either direction, precisely in viewof the very elusive and ambiguous
attitude of the author toward what is geometric and what is arithmetic in his treatment.
Thus, while the propositions of Book II are incorporated in Liber Mahameleth into an
arithmetic framework, the essentially geometric conception of the meaning of these
propositions is still very pervasive both in the rst, theoretical part of the treatise, and
in the second, more practical one. In particular, it is clear that the arithmetic ideas
developed here do not in any way embody any kind of algebraic implications.
5.3 Fibonacci
Leonardo Pisano Fibonacci (c. 1170c. 1250) is widely considered to be the most
prominent and emblematic gure in medieval European mathematics. He was well
acquainted with Greek, Byzantine, and Arabic sources and played a key role in com-
piling and helping disseminate in Western Europe innovative mathematical techniques
andconcepts, particularlythe use of the HinduArabic systemof numerationandmeth-
ods for solving equations. Fibonacci produced several works that reached relatively
broad audiences and in which he combined theoretical aspects with very practical tools
for land measurement and commercial calculations (Archibald 1913; Folkerts 2004).
In fact, he promoted a new kind of scientic genre, the practicae. These were brief
works, written in a popular style intended for lay readers, and with a clear emphasis
on applications (Hughes 2008, xviixxxv; Simi 2004).
It is not completely certain what version of the Elements was available to Fibonacci.
He may have simultaneously relied on several of the existing translations, including
the one made directly from Greek (Folkerts 2004, 109110; Hughes 2008, xix). At
any rate, there is no doubt that Fibonacci was well acquainted with Euclids text and
that he mastered all the techniques taught there. This is certainly the case for Book II,
whose propositions he cited and used explicitly as the geometric justication for the
problem-solving techniques that he borrowed fromIslamic sources. But it seems that it
was the Liber Embadorum, Plato of Tivolis Latin version of Bar H
.
iyya, which exerted
the most inuence on Fibonacci, particularly as manifest in De Practica Geometrie,
discussed below. Curiously, Fibonacci did not explicitly mention Bar H
.
iyya as Platos
source, perhaps because he considered the name Savasorda as a title, rather than as a
patronymic that might help identify the author (Hughes 2008, xxiiixxiv). Inuential
for him was also the work on algebra of Ab u K amil, which he may have known either
directly or indirectly.
1 3
678 L. Corry
Fibonaccis specic use of the propositions of Book II appears in Liber Abaci and in
De Practica Geometrie. Liber Abaci appeared rst in 1202 and then in a second edition
in 1228. This is perhaps the best known of Fibonaccis books and the main one through
which his inuence was felt over the next centuries in Europe, particularly in what
concerns the development of techniques for calculating and operating with square and
cubic roots, and for solving numerical problems. In Chapter 14, Fibonacci presented
some formulas or keys (claves), of fundamental importance for the techniques taught
in the book, comprising results which are clearly demonstrated in Euclids second
book. As the author of Liber Embadorum, Fibonacci did not provide geometric for-
mulations for these results, but just numerical examples. Of special importance were
those propositions corresponding to II.5 and II.6, since to them, Fibonacci explicitly
wrote, are reduced all the problems in algebra almuchabala (Sigler 2002, 490). It
is surely relevant for us, then, to examine how Fibonacci used these results within a
book of arithmetical content. But in order to grasp the context properly, it is helpful
to rst add some general comments about the book.
Geometric considerations appear in the book only after thirteen chapters of detailed
explanations about the decimal system and its uses for calculating and for solving all
kinds of numerical problems (but with nothing like symbolic notation or manipu-
lation). Still, it is clear that for Fibonacci such considerations were crucial from a
theoretical point of view because only the geometric interpretation could provide the
correct setting for handling surds. Thus, according to geometry, and not arithmetic,
the measure of any root of any number is found (Sigler 2002, 491). Fibonacci explic-
itly presented a procedure for nding such root: If you want to nd the root of 10, for
example, you need to nd two numbers that multiplied together make 10, say 2 and
5. You add them together and obtain 7, and you will order this to be the measure of
the line. This is done by arranging measured segments as in the following diagram:
Figure 25
Here abrepresents 2, andbc represents 5, andthe circle is tracedaroundthe midpoint
d, and eb is the perpendicular drawn on b. This segment is the root of 10, as is clearly
shown in geometry.
Fibonaccis attitude toward the results of Book II was along these lines. He did not
derive these results geometrically, and he continued to use, in the last chapters of the
book, a purely arithmetic language wherever the problems or the calculations involved
allowed for it. But he made it clear, in places where he relied on results of Book II, that
it is their origin in geometry that makes them useful as a source of legitimacy for the
arithmetical knowledge presented. The main place where this is done is in a section
1 3
devoted to the Solution of Certain Problems According to the Method of Algebra and
Almuchabala, Namely Proportion and Restoration. The cases treated by Fibonacci
are parallel to the six cases of al-Khw arizm. He presented themby combining numer-
ical examples, a formulation of the general rule (in purely rhetorical manner), and a
geometric justication of the procedure. Where necessary, Fibonacci used the Euclid-
ean results that he had already introduced, but without explicitly invoking them each
time. I illustrate this with the example of the case census plus a number will equal a
number of roots.
This case can be solved only if the number is equal or less than the
square of half of the roots, and if this is so, then the solution is
obtained as follows (Sigler 2002, 557):
[If] it is equal, then half of the number of roots is had for the roots
of the census, and if the number which with the census is equal to the
number of roots is less than the square of half of the number of roots
then you subtract the number from the square, and that which will
remain you subtract from half the number of roots; and if that which
will remain will not be the root of the sought census, then you add
that which you subtracted to the number from which you subtracted,
and you will have the sought census.
In the numerical example Fibonacci considered the question the cen-
sus plus 40 be equal to 14 roots. This is of course, the same case
we met in Bar H
.
iyyas book, but formulated without using subtraction
and of course using different numerical values. This case leads, as
we know, to two possible solutions (by subtracting or adding the said
square root), namely either 4 or 10. Fibonacci implied that for a
given problem only one of these values will be the solution sought
(the problems come later). What interests us here is, of course, how
this procedure is justified with the help of II.5, and this is seen
by referring to the figure that accompanies the text, which is the
following:
Figure 26
Schematically, the steps of the argument are the following:
(i.1) Take ab to be 14, and bisect ab at g. Also the point d cuts
ab, and we build a square dz, which is the census referred to
in the problem.
(i.2) The segment zi is traced equal in length to ab. Since zb is
the root of the census dz and since ab is 14, hence az, the
entire area, is 14 roots of the census, and hence the area ae
is 40, as stipulated in the problem.
(i.3) Now ae is ad times de, or, ad times db [remember, dz is a
square]. It is here that II.5 can be applied: ae =R(ad,db) +
Sq(gd) =Sq(gb).
1 3
680 L. Corry
(i.4) But gb is 7, as it is half of ab, and its square is 49. Hence,
as ae is 40, it follows that the square on gd is 9, gd is 3,
and db is 4, and this is the root of the sought census.
A slight variation in the argument and in the figure leads to the
second solution, 10. This proof is essentially the one already seen
in Bar H
.
iyyas text.
Notice that I used here the notation R(,) and Sq( ), only in step (i.3). This seems to
me to reect the way Fibonacci conceived the argument, namely an argument basically
on numbers, which at a certain points invokes a geometric result taken from Book II,
in order to reach the correct solution.
Again, to put things in the right context, it must be stressed that the section that
presents the six cases of solving quadratic equations (and where the above discussion
appears) is a relatively brief one. But the techniques introduced in this section are
subsequently used in the remaining chapters of the book, where Fibonacci discussed
a long series of specic arithmetic problems and showed how to solve them using
the techniques previously taught. Thus, we see how in Liber Abaci, a book which is
devoted to presenting a broad panorama of arithmetic and its applications, Book II is
used as a source of certain, geometric knowledge that is still needed for providing a
sound justication of the methods presented.
The second book of Fibonacci that interests us here is De Practica Geometriae,
composed in 1220 or 1221, between the two editions of the Liber Abaci. In many
respects the treatment of square and cubic roots, the references to Book II, and the
possible use of its propositions for solving problems involving squares of unknowns
are all quite similar to those of the Liber Abaci. But one important difference between
the two books is found in an entire section of the Practica where Fibonacci cited
twelve propositions from Euclids Elements, the rst nine of them being versions of
propositions in Book II (II.8 is absent). He also provided proofs for some of these
propositions which differ interestingly from Euclids or Herons. It is of interest for
us here to see the details of at least one of these proofs, and I will focus on II.9.
29
Proposition PG-38 corresponds to II.9 and the accompanying diagram
is the following (Hughes 2008, 30):
Figure 27
The line gd is divided into equal parts at point a and into unequal
parts at point b, and the proposition states that the squares on gb
and on bd are twice the two squares on da and on ab. The proof goes
as follows:
Since line ag was bisected at point b, the two squares on lines bg
and ba with twice the product of ab by bg equal the square on line
ga, that is, the square on line ad. Now the square on line ba with
just one product of ba by bg equals ba by ag. Whence the square bg
with the product of ba by ag and with that of ba by bg equals the
square on line ad. Let the square on line ba be commonly adjoined.
29
It would also be of particular interest to compare Fibonaccis proof of II.5 with all the others discussed
here, but in consideration with the length of this article I will focus on II.9 alone.
1 3
Then the two squares gb and ba with the products of ba by ga and ba
by bg are equal to the squares on lines ad and ab. But the square on
line ba with the product of ab by bg is as the product of ba by ag.
Therefore the square gb with twice the product of ba by ga is equal to
the two squared lines da and ab. When the square on line da with the
square on line ab exceeds the square on line gb by twice the product
of line ba by ga; that is, ba by ad. Likewise, because line bd was
bisected at point a, the squares on lines da and ab with twice the
product of ba by ad equal the tetragon or square on line bd. Therefore
tetragon bd exceeds tetragons da and ab by twice the product of ba by
ad. Therefore, by as much as tetragons da and ab exceed the tetragon
bg, by so much are they exceeded by tetragon bd. Whence tetragons
gb and bd are twice the squares da and ab.
Interpreting the steps of the proof (particularly the last one)
directly from the text is far from evident, so I render it schemati-
cally. What we get is the following:
(j.1) Since ga =gb +ba and ga =ad: Sq(ad) =Sq(ga) =Sq(gb) +SQ(ba) +
2 R(gb,ba)
(j.2) But, by construction, R(ga,ba) =Sq(ba) +R(gb,ba)
(j.3) Hence Sq(ad) =Sq(gb) +R(ga,ba) +R(gb,ba)
(j.4) Now, Sq(ad) +Sq(ba) =Sq(gb) +R(ga,ba) +R(gb,ba) +Sq(ba)
(j.5) Hence [by (j.2)], Sq(ad) +Sq(ba) =Sq(gb) +2 R(ga,ba)
(j.6) Or, equivalently: Sq(ad) +Sq(ba) =Sq(gb) +2 R(ad,ba)
(j.7) Now, since bd =ba +ad: Sq(bd) =Sq(ad) +SQ(ba) +2 R(ad,ba)
(j.8) Hence [from (j.6) and (j.7)]: Sq(bd) +Sq(gb) =2 [Sq(ad) +
Sq(ba)]
The conclusion in (j.8) is not transparent, and interpreting it with
the help of diagrams makes it easier. This can be done as follows:
1 3
682 L. Corry
Figure 28
If these diagrams do not really clarify how (j.8) actually follows
from (j.6) and (j.7), we can further try to explain this argument
by operating, in a proto-algebraic manner, with the various fig-
ures involved. Indeed, what Fibonacci says is that: by as much as
tetragons da and ab exceed the tetragon bg, by so much are they
1 3
exceeded by tetragon bd. Whence tetragons gb and bd are twice the
squares da and ab. Schematically, this is:
Sq(ad) +Sq(ba) = Sq(gb) +2 R(ad,ba) Sq(bd) = Sq(ad) +SQ(ba) +2 R(ad,ba)
And if we now subtract term by term the second identity from the
first, we obtain
[Sq(ad) +Sq(ba)] Sq(bd) = Sq(gb) [Sq(ad) +Sq(ba)] .
Hence:
2 [Sq(ad) +Sq(ba)] = Sq(gb) +Sq(bd)
as requested.
What has actually Fibonacci done here? As in the Liber Abaci, Fibonacci is build-
ing the legitimacy of his arithmetical techniques by reliance upon geometric results,
and by stressing the geometric character of the underlying arguments. But rather than
presenting the standard Euclidean proofs for those results taken from Book II, he has
preferred to introduce his own, new kinds of arguments. The latter lack the trans-
parency and simplicity characteristic of the Euclidean original, and they are also less
straightforward than Herons, whose operational approach Fibonacci essentially fol-
lows. But the underlying line of thought is a rather innovative and original one. Indeed,
Fibonacci has tried to incorporate into his geometric proofs techniques, or at least pat-
terns of thought, derived from the algebra almuchabala that he had come to master
so well. This is most clearly evident in step (j.8) above, and the way it was derived
from (j.6) to (j.7), as just explained. The kind of balancing of an equation that helps
understand the derivation is indeed the kind of balance and compensation technique
introduced in Islamic mathematics to reduce any given problemto the standard, known
ones. In this case, Fibonacci did not have an equation in the sense of an expression
with abstract symbols that could be operated upon, but he did have a situation simi-
lar to that found in the texts of Islam algebra, which rhetorically expressed relations
between numbers, unknown quantities, and their squares, and he used the techniques
of algebra almuchabala to simplify these expressions. Notice, moreover, that also
some other steps followed in the two proofs described above seem more easily under-
stood from this perspective. Thus for instance, (j.4)(j.5) nd a more natural place in
the arguments once we have conceded that Fibonacci followed techniques of algebra
almuchabala as part of his derivations.
It should be noticed that also the distributivity afforded by II.1 (or its direct deriv-
atives, such as II.2II.3) is crucial for all the arguments presented in De Practica
Geometriae. I discuss this point in some detail in [LC2].
This survey of the ideas of Book II as they appear in the works of Fibonacci affords a
further, interesting perspective on the rather original and divergent ways in which those
ideas could be handled in medieval mathematical treatises. Basically, Fibonacci con-
tinued to see in geometry a source of certainty which could account for situations where
arithmetic appeared as insufciently clear. At the same time, however, his way of using
geometric results on behalf of arithmetic was far fromuniform. He either took some of
the results for granted, or just illustrated them with numerical examples, or introduced
1 3
684 L. Corry
new proofs for the Euclidean results in which he tried an innovative, proto-algebraic
reasoning as part of the argument. This many-sided and rather exible viewof the rela-
tionship between geometry, arithmetic, and algebra certainly was conveyed, at least
at the implicit level, together with the more concrete results and techniques explicitly
taught in the treatises to the numerous readers of Fibonacci over the next centuries.
5.4 Jordanus Nemorarius
Jordanus Nemorarius is another highly interesting gure in Europe medieval math-
ematics and mechanics. Little is known of his life with certainty, but most likely he
lived in the early thirteenth century. Since Campanus mentions him in his version of
the Elements, Nemorarius surely lived before 1260. At the same time, from the con-
tents of his treatises, it is clear that these were written after the various twelfth-century
translations of the Elements. From a variety of scientic topics on which he wrote
original treatises, his two most important arithmetical texts are De elementis arismet-
ice artis and De numeris datis. Some of the results in De elementis are applied in De
numeris datis, which indicates that the former work was composed before the latter
one (Busard 1992, 123). It is also of interest to mention that there were failed attempts
to produce printed versions of these two works by both Regiomontanus (14361476)
and Francesco Maurolico (14941575). Finally, only the Arithmetica (as De elementis
came to be known) appeared in two printed editions of 1496 and 1514 by the French
humanist Jacques Lefvre dtaples (14551536), also known as Stapulensis (Lefvre
dtaples 1946). In the early sixteenth century, this book continued to be used as a
textbook in many European universities (Busard 1991, 711). Likewise, even though
De numeris Datis was never printed, a large amount of extant copies, dating from
the thirteenth and up to the sixteenth centuries and found all around Europe, testify
to a continued interest among mathematical audiences (Hughes 1981, 2021). Still,
it seems that the inuence of Jordanus on his contemporaries was relatively limited,
possibly because the mathematical content of this work was deeper than, and deviated
in style from, the more didactical and metamathematically oriented texts associated
with the Adelard-Robert tradition (Hyrup 1988, 341343). Campanus de Novara,
however, did read Jordanus and he incorporated many of his ideas into his version
of the Elements, as we shall see right below, and in this way, Jordanus ideas could
indirectly be of great inuence for the processes discussed here.
Our focus here is on Jordanus Arithmetica, a treatise that embodies an attempt to
provide a comprehensive, and indeed pioneering, theoretical foundation for arithmetic.
Many details of the presentation seem to derive from a conscious striving to develop
arithmetic without relying on geometrical concepts or results of any kind, while at
the same time retaining the model and the standards of rigor set for geometry in
Euclids Elements. In doing so, it comprised a rather exhaustive overview of the body
of theoretical Euclidean and Boethian arithmetic as was known at the time (though,
curiously, the Elements are never explicitly referred to in the text). The treatise also
betrays the inuence of al-Khw arizm, Ab u K amil, and Leonardo (Hyrup 1988,
310), and it is quite evident that al-Nayrzs text may have been among Nemorarius
main sources. In addition, in matters of overall style, if not of detail, the inuence of
1 3
Liber Mahameleth seems to surface at various places in the text. Yet, the Arithmetica
was more than just a straightforward compilation of known results, and not only
because it included many new propositions and new proofs of known propositions,
but also because the style was highly original and because the synthesis thus produced
presented a new overall image of a discipline that might have been systematically
studied and further developed.
30
Nemorarius presentation starts from fourteen denitions, which somewhat over-
lap, but are not identical with, those appearing in Book VII of the Elements. He also
included three original postulates (petitiones) and eight axioms (communes animi
conceptiones), from which he derived more than four hundred propositions organized
in ten books (the total number of propositions varying across the different extant
manuscripts. See Busard 1991, 12). As in the Adelard II version of the Elements, in
Nemorarius text we often nd a sketch of the argument that the reader is supposed to
complete. Unlike Boethius, Jordanus did not illustrate his arithmetic arguments with
numerical examples.
31
And unlike Euclid, only some of the propositions are accom-
panied by diagrams where lines represent the numbers. This is a rather noteworthy
point. In simpler proofs, where few different numbers are referred to and little con-
fusion is likely to arise, Jordanus always preferred to refer to the various numbers
involved in the argument using expressions such as the rst number, the largest of
the two, the product of the whole by each part, etc., and where possible he avoided
calling them by letters. It is only when the argument became somewhat more involved
that he introduced letters that served as names for the numbers referred to in the
proofs. Only in such cases did he add letters to the diagram to indicate the additional,
named numbers (as we shall see in an example below). And of course, whatever letters
appeared in the proofs, Jordanus never operated on them as on abstract symbols. In
this sense, Arithmetica displays noteworthy similarities to Liber Mahameleth. And,
as in that book, propositions from Book II (namely II.1II.5), or variants thereof, play
an important role here in creating the theoretical foundation on which Jordanus pur-
ported to build the entire edice of arithmetical knowledge. This is particularly the
case concerning distributivity, for which Jordanus presented a detailed and original
treatment in the preliminary section of the book, while drawing, wherever possible,
only on the denitions, the postulates, and the axioms presented at the beginning.
I discuss this important issue in detail in [LC2].
Here, the focus is on Jordanus rendering of II.5 in the Arithmetica (which I call here
A-I.19), which offers many interesting vistas to his overall approach. It is important,
rst of all, to explain the context where it appears as part of the preliminary section.
After treating basic properties such as commutativity and distributivity in its various
cases, Jordanus discusses Euclids II.4 and several variants of it, namely general rules
for separating a given number into two arbitrary parts and then nding the relation
30
(Hyrup 2010, 16) has argued that the potential impact of the Arithmetica materialized only to a very
limited extent. Here I want to focus on the way in which, via its reading by Campanus, it did have an
important inuence in the specic issue of the increasingly arithmetic-algebraic conceptions of Book II.
31
The Stapulensis printed version of 1496 does include numerical examples. In this as well as in other
respects, it is interesting to compare the printed with the original Jordanus version. We shall leave this
comparison for a future opportunity.
1 3
686 L. Corry
between the square of the number and the partial products of its parts (this is discussed
in detail in [LC2]). Then come two propositions that take this situation one step further,
namely separating a given number in two parts, but now in two different ways, and
examining the relations between the possible partial products that arise thereby. The
two propositions are formulated as follows (Busard 1991, 70):
A.I-19: If a number is divided into two equal and into two unequal parts, then the
product of one of the equals by itself equals the product of one of the unequals by
the remaining one together with the product of the difference by the difference.
A-I.20: If a number is divided into two parts in two ways whatsoever, the product of
the greatest of the parts by the smallest together with the product of the difference
of the [larger one] and one of the others by the difference of the same and the
smaller equal the product of the two intermediate numbers.
The accompanying diagrams of the propositions are, respectively, as follows:
Figure 29
In the rst case, we are told that a and b are the greater and lesser portions respec-
tively, whereas c is one of the two equal ones, both of which differ equally from
b. The difference is called d, and the proposition proves that the product of a by b
together with that of d by itself equals the product of c by itself. In the second case,
we are told that a and b are the greater and lesser portions, respectively, whereas c and
d are, respectively, the greater intermediate and lesser intermediate ones. Here, e is the
difference between a and c, while f is the difference between a and d. The proposition
thus proves that the product of c by d equals the product of a by b together with that
of e and f . Clearly, proposition A-I.20 is a generalization of A-I.19, for the case
where also the second separation is into unequal parts.
32
This becomes more clearly
apparent if we render them symbolically as follows:
A-I.19: If a +b = c +c, then a b +(a c) (a c) = c c
A-I.20: If a +b = c +d, with a > c > d > b, then a b +(a c) (a d) = c d
A closer look at the proofs brings to the fore some additional points that are of the
foremost importance for a clearer understanding of Jordanus, and the various kinds of
32
It must be stressed, however, that Jordanus included in his treatise another proposition, A-X.3, which
is in fact equivalent to A-I.19 and that embodies the so-called Regula Nicomachi: given three numbers in
arithmetic progression, a b = b c, then we have b
2
a c = (a b)
2
. Boethius has stated in his
Arithmetica that this rule was discovered by Nicomachus (Busard 1991, 14). Jordanus very short proof of
A-X.3 simply invokes A-I.19. This apparent repetition reects, so it seems to me, Jordanus conscious view
of A-I.19 as naturally belonging to the context of this preliminary type of results, where a given number is
separated in two parts and the relations between the partial products, even if an equivalent form of it could
appear in the context of a later book as well.
1 3
ideas involved in his treatise. For convenience I have rendered here the proofs schemat-
ically, without directly quoting them. This may give rise to some misrepresentation
which I will nevertheless strive to avoid, among other things by translating word by
word Jordanus rhetoric, while even retaining his own symbols exactly the way they
are used in the argument.
The steps of the proof of A-I.19 are the following (square brackets
enclose comments that do not appear in the original):
(k.1) a +b = c +c; d = c b [or c = d +b]
(k.2) By A-I.15 [i.e., Euclid II.4]: c
2
= (d +b)
2
= d
2
+b
2
+2 b d
(k.3) But 2d = ab [or b +2d = a (but this is not explicitly said!)]
(k.4) Hence by A-I.14 [i.e., Euclid II.3]: b
2
+2 b d = b a
(k.5) Finally [from (k.2) to (k.4)]: b a +d
2
= c
2
In the proof of A-I.20 Jordanus invokes some additional results such

as A-I.3, A-I.8 (commutativity of the product), and A-I.9 (distribu-
tivity). I do not comment here directly on these three propositions,
as I say more about them in [LC2]. The steps of the proof of A-I.20
are the following:
(k.6) By A-I.3: since a +b = c +d, then, a c = d b; and a d = c b
(k.7) Hence, by A-I.9: c d = c b +c e
(k.8) But by A-I.8: c b +c e = b c +e c
(k.9) Again, by A-I.9: e c = e b +e f
(k.10) But e b = b e, and also, by A-I.9, b c +b e = b a
(k.11) Finally, c d = b a +e f
Let me now draw attention to the following, remarkable point: the
crucial step in the proof of A-I.19 is (k. 3): since twice d is the
difference of a and b. Curiously, however, Nemorarius simply states
this relation and provides no explicit justification for it. How is
then this step justifiable? On first sight, one may be tempted to see
this identity as following from (k.1), via some kind of symbol manip-
ulation. This would involve something like the following putative
steps:
(k
.1) a +b = 2 (d +b) a = b +2 d 2 d = a b
But notice that what is actually used in step (k.4) is the relation
b + 2 d = a, which is the midstep in this putative derivation and
which I indicated in square brackets in (k.3). This would make the
relation 2 d = ab irrelevant for the derivation. Why would Jordanus
then write twice d is the difference between b and a rather than
a equals b and twice d, which is the rhetorical counterpart of
the relation needed for step (k.4)? It seems, therefore, that we
must look for the justification elsewhere, and as Jordanus remains
silent, we can only speculate. My guess is that the sentence twice d
is the difference between b and a refers to a situation that can be
derived from a visual inspection of the diagram. This can be seen by
imaginarily extending in the following way the situation described
in the original diagram of A-I.19:
Figure 30
1 3
688 L. Corry
By directly inspecting a diagram like this one, it becomes clear
that, indeed, twice d is the difference between b and a, as stated
by Jordanus.
We can find a rather similar situation in relation with the proof of
A-I.20. Let me first stress how the concluding step (k.11) is actually
reached. From (k.7) to (k.8), one obtains c d = b c + e c, and then,
using (k.9) we get c d = b c+e b+e f. From here, using (k.10) we get
the desired result c d = b a +e f. But notice that Jordanus invokes
A-I.9 in three of the steps, whereas in fact, in order to apply this
result in each of the steps some additional relations are required
that are not explicitly mentioned in the text. Thus, for instance,
how do we know in (k.7) that d = b + e, in order to conclude that
c d = c b +c e? Or how do we know in (k.9) that c = b +f, in order to
conclude that e c = e b+e f? Like in the case of A-I.19, we might, on
first sight, suggest putative symbolic derivations that would help
explain how Jordanus obtained the said relations, this time starting
from (k.6). Thus, for instance, we might think of the following two:
(k
.2) a c = e(by def.), but a c = d b; hence d b = e and d = b +e

(k
.3) a d = f(by def.), but a d = c b; hence c b = f and c = b +f

But the problem with such an assumption is that not only does Jordanus
remain silent about any derivation of this kind, but also that we do
not find any other similar one, as far as I can see, in other parts in
his treatise. Thus, I suggest that my conjecture for A-I.19 will also
work in this case, if we imaginarily adapt, in the following way, the
situation described in the original diagram of A-I.20 (and which is
not much different from what I suggested above for A-I.19):
Figure 31
As above, by directly inspecting these two diagrams, it becomes clear
that, indeed, d=b+e and c=b+f, as implicitly used by Jordanus. More-
over, from the left-hand side, it also becomes clear that a=c+e, a
relation which is implicitly used in (k.10).
These two proofs illustrate important characteristics of Jordanus approach. On the
one hand, he goes to great pains in order to set clear the foundations of arithmetic and
to abide by the standards set by Euclid in the case of geometry. On the other hand, he
seems to have implicitly adopted some results that are necessary for completing the
argument, without even commenting on them. I think that my conjecture above, about
his possible use of diagrams as the source for these implicit assumptions, sheds some
light on the situation. We already saw above several instances of implicit reliance on
diagrams embedded in proofs. Thus was the case with Euclids use of distributivity
in the case of visible gures, or with some proofs of al-Khw arizm and Ab u K amil.
But what is of special interest in the case of Jordanus is that, unlike those mentioned
previously, he seems to have been relying on properties of diagrams of the arithmetic
1 3
type, i.e., where the lines are not part of a geometric construction but rather, just
indicative of the numbers involved.
33
I have also stressed this point by schematically
rendering the proofs without reference to geometric constructions using R(,) and Sq( )
as with other authors above. But beyond the validity of my conjecture about Jordanus
reliance on diagrams, it can be said with certainty that his arguments do not involve
any kind of techniques that merit the adjective algebraic. The proofs are not based
on abstract manipulation of symbols. They do not even involve following rules of
al-jabr and al-muq abala, as Fibonacci did in his proofs mentioned above. And yet
at the same time, all the preliminary section of his treatise is devoted to developing
general rules of arithmetic that can be applied in changing situations. This is indeed
what makes the book interesting as part of our account here, since it shows the extent
to which a foundation of arithmetic and an examination of the properties of the basic
rules of calculating with numbers could be pursued without thereby moving into the
territory of algebra. Specically for II.5, it is interesting how Jordanus has moved it
into a completely different realm from where it was originally conceived, and it was
placed, in a truly natural way, within the framework of a rather thorough exploration
of general, foundational rules of arithmetic. It is no wonder, then, that anyone who
was exposed to this kind of treatment of II.5 could read this proposition, as part of
Book II, as being about arithmetic (and in later periods even as being about rules of
algebra). Some of this is already apparent in Campanus de Novara as we shall now
see.
5.5 Campanus
I want to consider nowthe thirteenth-century Latin version of Euclids Elements due to
Campanus of Novara (c. 12201296), completedsometime between1255and1259. As
already said, this is the text that dominated European mathematics until the sixteenth
century, and more than one hundred manuscripts of it have survived (Folkerts 1989,
38-43). Like other earlier translators and commentators discussed above, in preparing
his text Campanus was also guided by clearly dened didactic considerations. He made
considerable efforts to present Euclids text as self-contained as possible and free from
obscure or debatable points such as found in Roberts text (Busard 2005, 32). Indeed,
Campanus repeated many of Roberts enunciations, but at the same time many of his
proofs were original. More importantly, he also added signicant amounts of material
taken from sources such as al-Nayrzs commentary and Jordanus Arithmetica. This
is particularly remarkable in the case of Book IX, where Campanus incorporated into
the original Euclidean text thirteen additions and comments comprising arithmetic
versions of propositions from Book II. This is one of the main aspects that marks a
clear difference between Campanus version of the Elements and those that preceded
it. Clearly all of this is highly relevant for our discussion here.
33
It is pertinent to refer the reader in this regard to (Puig 1994). This seldom-cited, but highly original
article makes a remarkable connection between the arithmetic results presented in Jordanus De numeris
Datis and the diagrammatic aspects underlying the arguments of the proofs. I discuss this in some detail in
[LC2].
1 3
690 L. Corry
These additions are an interesting manifestation of the broader issue of Campanus
attitude toward the relationship between arithmetic and geometry. This relationship,
in turn, is related to a more general medieval concern with the existence of parallel
concepts and propositions in the Elements for handling separatedly magnitudes and
numbers. In earlier versions of the Adelard-Robert tradition, this concern was sug-
gested in several places (Murdoch 1968, 8889). In Campanus, it becomes more
explicit and central, and his additions and comments reect an active interest in trying
to come to terms in original ways with this concern.
Thus for instance, in his comment to VII.12 (which corresponds to Euclids VII.11),
Campanus asked why Euclid proved the more particular case of proportions with
numbers in Book VII, if the general case for magnitudes had already been proved in
Book V. The reason he gave for this is that the propria principia of the two books
are different and, hence, corresponding propositions should be proved separately and
based on those principles alone in each case. In particular, he stressed, as the principles
of Book VII relate to numbers, they are closer to the intellect and hence are more
easily comprehensible than those treated in Book V, affected as the latter are by
the wickedness of the incommensurable quantities (propter malitiam quantitatum
incommunicatum) (Busard 2005, 240; Rommevaux 1999, 8992).
34
Against this background, let us consider more specically the way in which Cam-
panus presented the ideas of Book II. In the rst place, I note that in Book II itself
Campanus generally followed Euclids original approach and thus remained within a
completely geometric context. For propositions II.2II.4, moreover, Campanus also
presented alternative proofs along the lines of Herons approach (which he proba-
bly learnt from al-Nayrzs commentary). The most interesting sentence in the entire
Book II, however, appears at the end of the proof of II.10. Clearly intended to mark a
separation between the rst ten and the last four propositions of the book, Campanus
stated that this and all the previous propositions, are true for numbers as well as for
lines (Busard 2005, 103: hec autem et omnes premise veritatem habent in numeris
sicut in lineis). This statement clearly allowed and justied using the rst ten results
of Book II in the context of the three arithmetic books of the Elements, as we shall see
now.
Campanuss treatment of Euclids three arithmetic books contains many modi-
cations, additions, and comments to the original propositions.
35
In Book VII, for
instance, he added numerical values to lines in the diagrams that accompany several
of the propositions, apparently as a means to helping the reader gure out the situation
more easily.
36
In addition, in some propositions unity is explicitly indicated in the
gure either with its value 1 or with the word unitas. None of the numerical values,
however, is used or mentioned in any way in the proofs of those propositions.
But as already stated, the most interesting additions appear as thirteen comments
to a proposition in Book IX, namely IX.16, which Campanus took from Jordanus
34
See also Campanus comments to VII.6 (p. 236), which is compared with V.13 (i.e., Euclid V.12).
35
All of these are very interesting in themselves, but discussing them would be beyond the scope of this
article. See (Rommevaux 1999, 93100).
36
The only other place where we nd such additions to the diagramis in the very last arithmetic proposition,
IX.39.
1 3
Arithmetica.
37
Eight of these comments are arithmetic formulations of propositions of
Book II,
38
but they do not comprise numerical examples such as found in al-Nayrzs
text. Campanus provided proofs for all of them, and they are highly original and
interesting. In those proofs that concern properties of distributivity (i.e., those related
to II.2II.4), we nd further evidence of Campanus focus on the relationship between
Books V and Books VII (I discuss this in some detail in [LC2]). Here, I want to focus
on the proof of II.5, whose argument substantially deviates from that of either Heron,
or al-Nayrz or Jordanus.
39
What is of particular interest is that Campanus proof is
fully arithmetic not only in the sense that it is evidently formulated as a property of
numbers, but also in the sense that he uses an argument which is unlikely to have
arisen in the kind of geometric context that Euclid originally devised for II.5. Let us
see the details.
Campanus arithmetic version of II.5 appears as comment 7 of IX.16.
The proof relies on comment 6, which is equivalent to Euclids II.4.
The accompanying diagram is the following (p. 293):
Figure 32
As usual, the segment ab is divided in equal parts at c and in unequal
parts at d, and the argument of the proof reads as follows:
By the previous statement [i.e., II.4], the square of cb equals the
square of cd and the square of db and twice the product of bd in cd.
The products of bd by itself and by cb equal that by cb by the first
comment here, and hence it equals that by ac. Hence, the product of
bd by itself and twice that by cd equals that of bd by ad. And for the
same reason the square of cb exceeds by the square of cd the product
of bd by ad.
Schematically, the argument is the following:
(l.1) By II.4: cb
2
= cd
2
+db
2
+2 db cd
(l.2) By II.3: bd
2
+bd cd = bd cb
(l.3) But cb = ac; hence bd
2
+bd cd = bd ac
(l.4) Hence: bd
2
+2 bd cd = bd ac +bd cd = bd ad
(l.5) Hence, by (l.1) and (l.4): cb
2
= cd
2
+bd ad
The step that requires special attention here is (l.4). It is based on a tacit addition of
bdcd to both sides of an identity, followed by an application of II.3. This is the step that
betrays the purely arithmetic context in which Campanus conceived the argument of
his proof. Indeed, it would be highly unnatural, and certainly alien to Euclids practice,
to imagine a diagram that could graphically represent the passage from (l.3) to (l.4),
that is, from bd
2
+bd cd = bd ac to bd
2
+2 bd cd = bd ac +bd cd. Because
37
See (Busard 2005, 291), and also, proposition A-IV.19 of Jordanus in (Busard 1991, 103). Campanus
proof is somewhat different from (and more detailed than) Jordanus.
38
These are comments 412. Comments 13 correspond to Jordanus A-I.9A-I.11, whereas comment 13
corresponds to Euclids V.11. See (Busard 2005, 33).
39
The difference is directly noticeable by comparing steps (l.1)(l.5) here below with steps (c.1)(c.7) for
Heron, with (k.1)(k.5) for Jordanus, or with the algebraic interpretation that (Curtze 1899, 96) provides in
a footnote for Al-Nayrz.
1 3
692 L. Corry
of the squares, it is even difcult to imagine how this derivation could be represented
in an arithmetic diagram of the kind that I conjectured above for Jordanus argument
(and of course I have not rendered here the products as rectangle formation). Nor is
it natural, I think, to interpret this kind of reasoning as embodying ideas derived from
algebra almuchabala, as we have seen above for the case of Fibonacci. Rather, the
passage from (l.3) to (l.4) is more naturally seen, I think, as arising from a direct
application of the kind of general arithmetic rules that Jordanus worked out in his
Arithmetica, and the kind of possible manipulations of numbers arising from it.
40
At this point, we have arrived at one of the most signicant milestones of this histor-
ical journey, given the decisive inuence of Campanus in the process of transmission of
the Elements to the printed world of the European renaissance and its subsequent dis-
semination. A typical reader of this version of the Elements would become acquainted
with Campanus discussion about the double presentation of the theory of proportions,
once for numbers and once for general magnitudes, and about the different nature of
these two kinds of entities. Then, he would confront the results of Book II presented
as comments to a result of Book IX, and bearing a purely arithmetic character in
both their enunciations and the various steps of the proofs. And at the same time, of
course, he would also read the same results in their fully geometric presentation, in the
framework of Book II. With these two versions in front of him and with the increas-
ing attention devoted to algebraic methods from the sixteenth century on, the reader
might lay the stress on the essential differences between the two ways of presenting
the same results as pertaining to two essential separate domains (one arithmetic and
one geometric), but he might as well conate both presentations as expressing two
faces of one and the same mathematical idea. Thus, for a historian trying to understand
the ways in which the results of Book II were typically read and understood on the
wake of the Campanus version, the assumption that these results were conceived in
arithmetic or algebraic terms, rather than as purely geometric ones, will not involve
anymore the kind of anachronistic interpretation that was involved in making a simi-
lar judgment about the way in which the same results were conceived at the time of
Euclid, and even by most other later authors whose works we have been investigating
thus far. After Campanus in a denite manner, and perhaps also somewhat earlier in
less clear-cut ways, the possibility of interpreting the results of Book II in terms of
general rules for operating with numbers did not require the reader to incorporate into
the text any mathematical idea not itself contained in the very text of the Elements
that was available to him. At the same time, as we shall see now, not every reader of
the Elements, posterior to Campanus, would necessarily adopt an entirely arithmetic
reading of these results, and geometry was not quickly discarded as the source of
certainty and legitimacy to be relied upon.
40
Though I would not like to lay excessive stress on names, I feel it is improper to claim, as in (Busard
2005, 39), that the algebraic method which Campanus used for proving Campanus IX.16 add. 412 agrees
with the method which Anairitius used for proving II.2II.10. For one thing, I already indicated that both
the method and the line of argumentation actually differ from that of Al-Nayrz. But also, using the
term algebra in this context may be misleading since nothing here is manipulation of abstract symbols
according to formal rules, not even in the manner intended in Islamic mathematics. Rather, Campanus
simply operates on numbers (which are indicated with letters that serve as names) according to general,
arithmetic rules.
1 3
5.6 Gersonides
Levy Ben Gerson (12881344), also known as Gersonides, was the most prominent
gure in the tradition of Jewish medieval sages with a keen interest in mathematics.
If Bar H
.
iyya was the initiator of this tradition, Gersonides was certainly the rst to
contribute truly original mathematical ideas. He wrote treatises in various elds of
interest, scientic as well as of Biblical exegesis. The most mathematically interesting
of them, dating from1321, is Sefer Maaseh Hoshev ( ), a title that can be
translated as The Work of the Calculator. Gersonides also wrote a commentary on
Books IIV of the Elements and a separate treatise on geometry, from which one may
perhaps infer the extent and sources of his knowledge. Thus, as Tony Lvy has carefully
documented (Lvy 1992, 8792), there are some similarities between Gersonides text
and some of the ideas found in al-Nayrzs commentary. Gersonides may have been
acquainted with the latter either in its Arabic original or in Gerard of Cremonas Latin
translation, or perhaps also via the Campanus edition. Still, other extant manuscripts
raise the possibility of alternative sources and ways of transmission of Euclids texts
to Jewish mathematicians. In his commentary on the Elements, Gersonides dealt with
purely geometric questions such as a possible proof of the fth postulate. Of Book
II, he only discussed Proposition II.13. However, arithmetic versions of propositions
from Book II do appear in Maaseh Hoshev and they are of interest for us here.
Maaseh Hoshev is a book on arithmetic and combinatorics in two parts.
41
The
Hebrew wording of the title embodies a thoughtful word play which opposes two
terms: one stressing the practical aspect of the treatise ( ) and the other stressing
the theoretical one ( ). Gersonides explained in the introduction that the purpose
of his treatise was not just to teach the practical rules for solving arithmetic problems,
but also the theory underlying those rules, since mastery of the former necessitates a
full understanding of the latter. Gersonides also suggested that the prospective reader
of his book should rst master the contents of the three arithmetic books of the Ele-
ments, since proofs for propositions in those books would not be repeated the text. But
the image of arithmetic that Gersonides presented was much broader than what the
Elements contained, and some of the proofs in the text go well beyond Euclid. Maaseh
Hoshev, for instance, discussed many additive properties of the natural numbers as
well as combinatoric results. He also followed ingenious arguments that came very
close to mathematical induction (Rabinovitch 1970).
Gersonides started by proving some elementary properties of addition and multi-
plication, in a manner that reminds us of the preliminary section of Jordanus Arith-
metica discussed above. This preliminary section comprises results on distributiv-
ity and associativity of the product, as well as on identities involving the squares
of a number separated into two parts. But Gersonides was much less systematic or
exhaustive than Jordanus in his treatment. He included propositions that would be
needed for proving more complex propositions later on in the text, as well as some
41
Here, I refer to the text as is appears in (Lange 1909), which is where the Hebrew text rst appeared in
print in modern times, together with a German translation. See also (Simonson 2000a,b) for a description
of the contents, for some important remarks concerning the existing manuscripts and versions, and for
additional parts not appearing in the Lange edition.
1 3
694 L. Corry
others that are not even used, but it is not easy to discern a clear criterion for this
choice.
Arithmetic versions of the rst six propositions of Euclids Book II appear among
these preliminary results. Propositions II.1II.4 are proved by straightforward arith-
metic argument, namely by counting units in each case. (I say more on this in [LC2].)
But when we read Gersonides versions of II.5II.6 (which are, respectively MH-I.8
and MH-I.5), we nd a more interesting and complex picture. Let us start by looking
at Gersonides text for MH-I.8, which reads as follows:
42
The area obtained by multiplying half of a given number by itself equals the
area obtained by multiplying a part of that number by the remainder together
with the square of the difference between the part and half of the given number.
Therefore let AB be the given number and let it be halved at point C, and divided
at some other point D. I say that the square on the number AC equals the area
obtained by multiplying the number AD by DB together with the square on CD.
Figure 33
Proof: the square on AC equals the area of AC on CD together with the
area of AC on DB, but the area of AD on DB equals the area of AC on DB
together with the area of CD on DB. If we subtract the area of AC on
DB which is common to them, then what remains from square AC equals
the area of AC on CD, which equals the area of CB on CD, whereas what
remains from the area AD on DB is the area CD on DB. But the excess
of the area CB on CD over the area CD on DB equals the square on CD.
Therefore, the square on AC equals the area of AD on DB together with
the square on CD, and this is what we wanted.
Like Heron, also Gersonides bases the proof on using the elementary distributivity
results of Book II in order to decompose, transform, and then compose again, the basic
gures obtained. He may have learnt this approach from al-Nayrz. But besides this
basic similarity, his argument evidently takes a longer and rather less transparent path.
This argument is more clearly understood when rendered symbolically, as follows:
(m.1) Sq(AC) = R(AC,CD) +R(AC, DB) [by II.2]
(m.2) On the other hand, R(AD,DB) = R(AC,DB) +R(CD,DB)
(m.3) Hence Sq(AC) R(AC,CD) = R(AC,DB) = R(AD,DB) R(CD,DB)
(m.4) But AC = CB [and hence R(AC,CD) = R(CB, CD)]
(m.5) Hence Sq(AC) R(CB, CD) = R(AC,DB) = R(AD,DB) R(CD,DB)
(m.6) But R(CB,CD) = R(BD,CD) + Sq(CD) [by II.1] [and therefore Sq(AC)
R(BD,CD) Sq(CD) = R(AD,DB) R(CD,DB)]
(m.7) Hence Sq(AC) = R(AD,DB) +Sq(CD),
42
This is the version appearing in (Lange 1909, 5). It is remarkable that the proposition was not included
in the second edition. See (Simonson 2000a,b, 297).
1 3
If we wanted now to visualize the argument in terms of a diagram we
could start with steps (m.1)(m.2) as follows:
Figure 34
Step (m.3) introduces a kind of derivation that we had not seen so
far, and in which the two above figures are compared by way of sub-
traction of areas, as follows:
Figure 35
In step (m.4) the equality CB = AC is invoked to obtain the identity
in step (m.5):
Figure 36
This step implies going even further than in step (m.3) since, unlike
there, the areas that are subtracted here are not in the given figures
to begin with. A further decomposition in (m.6) allows reaching in
step (m.7) the identity to be established:
1 3
696 L. Corry
Figure 37
As with other cases above, also in this case one might raise the question whether
this geometric rendering of the text is the correct one, or whether we should rather
follow an arithmetic rendering. That is, we might ask whether Gersonides conceived
the argument while thinking about construction and manipulation of geometric gures
as suggested in the diagrams above, or whether, on the contrary, he thought in terms
of general rules of multiplication and addition for numbers. The inclusions of steps
such as (m.3) and (m.5) make this question of particular interest. I think that in this
case there are good reasons to attribute Gersonides a more geometric kind of thinking,
at least for this specic proposition. One reason for this is the language he used, in
which a product is always dubbed area ( ) or square ( ). This is of course
inconclusive evidence, and we saw above some examples to the contrary, but it is at
least indicative. Also, he did not choose here the arithmetic counting of units that he
followed in other cases. But a more important consideration is that, while we have
not seen other examples where areas are subtracted as part of a geometric argument,
we have not seen either any example where this is done for numbers as part of an
1 3
arithmetic one. And indeed, I nd it unnatural to see how some of the crucial steps
of the argument could have been conceived as part of a purely arithmetic derivation.
Thus, for instance, it is not clear what could be the arithmetic idea behind a putative
arithmetic rendering such as the following:
.3
AC
2
AC CD = AC DB = AD DB CD DB
Moreover, the geometric rendering also seems appropriate for the argument of MH-I.5,
whose proof is based on repeated applications of distributivity properties, and which
is accompanied by the following diagram:
Figure 38
Here, the number AB (Gersonidess term) is bisected at C and the number BD
is added. The proposition then states, as in II.6, that the area of AD by DB together
with the square on CB equals the square on CD.
Gersonides proof is as follows:
(n.1) R(AD,BD) = R(CD,DB) +R(AC,DB)
(n.2) But R(AC,DB) = R(BC,DB)
(n.3) [Hence R(AD,BD) = R(CD,DB) +R(BC,DB)]
(n.4) Hence by adding Sq(CB),
R(AD,DB) +Sq(CB) = R(CD,DB) +R(BC,DB) +Sq(CB)
(n.5) But Sq(CD) = R(CD,BD) +R(CD,CB)
(n.6) But R(CD,CB) = R(CB,BD) +Sq(CB)
(n.7) Hence Sq(CD) = R(CB,BD) +R(CB,BD) +Sq(CB)
(n.8) Hence Sq(CD) = R(AD,BD) +Sq(CB),
In this case we find no subtractions as in the previous example, but
here the steps are more straightforward and one can simply imagine
them as being similar to those followed by Heron in his operational
geometric proof.
Summarizing this section, then, one can say that Gersonides version of these two
propositions is not easily classied as either geometric or arithmetic. On the one hand,
it is clear that for him the two propositions are meant (like all other propositions in
the book) to express properties of numbers, and not of geometric gures. On the other
hand, while for other propositions in the book which were also arithmetic versions
of propositions in Book II, he had provided arithmetic proofs, here he proceeded in
ways that remind one of those of al-Khw arizm and of his Islamic followers; that
is, he provided a geometric justication for a result related to numbers. It is evident
that lacking a exible language in which the various steps of the proof (as stated
above) could be conceived and formulated in arithmetical, or proto-algebraic terms,
an operational approach to geometric manipulation of areas, such as implicit in Herons
proof, offered a blueprint of a proof that could be more easily adapted to the arithmetic
spirit of the book. Euclids proof would have also worked here, of course, but it would
have been much less akin to the arithmetic spirit of the proposition as conceived by
1 3
698 L. Corry
Gersonides. In this sense, Gersonides proofs do reect a well-developed ability to
manipulate abstract relations between numbers, albeit without having at hand a fully
developed symbolic language. Of course, the proofs discussed here are not among the
most complex ones that Gersonides handles in the treatise, and yet I think that this
conclusion applies broadly beyond the specic cases considered. In terms of historical
development, at any rate, since Maaseh Hoshev had little or no visible inuence on
later mathematical developments in Europe, this highly original version of II.5 soon
fell into oblivion.
5.7 Barlaam
The last text that I want to consider here is a highly original collection of results
embodying arithmetic versions (not a commentary) of Euclids propositions II.1
II.10, and commonly attributed to the fourteenth-century Basilian monk and scholar
Barlaam de Seminara (ca. 12901348).
43
Barlaam is mentioned in passing in Heaths
edition of the Elements (Heath 1956 [1908], Vol. 1, 74), as well as in some other places
in the secondary literature, but it seems that there exists no serious historical research
on him and on his startling Euclidean text.
In the text, we nd no preliminary explanations about the background, or about what
exactly the author had in mind when composing it. Rather, the text itself opens with
arithmetical denitions similar in spirit and in wording to those found in Book VII:
multiples of numbers, plane numbers, parts of numbers. More generally, the wording of
the propositions and, especially, the accompanying diagrams resemble those found in
Books VIIIX: They are not geometric constructions needed to support the argument,
but rather indications, with the help of various lines drawn one next to the other, of
the various numbers mentioned in the proof. Still, in some places Barlaams use of the
lines appearing in the diagram deviates in important senses from standard usage. Let
us see the details of his proof of II.5.
Barlaams proof of this proposition relies on the use of II.1, and
also, in an original way, of II.4. The accompanying diagram is as
below (Heiberg and Menge 18831893, Vol. 5, 730732):
Figure 39
The standard line appearing in all diagrams of proofs of II.5 also
appears here, with ab representing a number (Barlaam stresses that
this is an even number), that is divided into two equal numbers ag,gb
43
The full original Greek text appears in (Heiberg and Menge 18831893, Vol. 5), as Appendix Scholiorum
4. I thank Michael Fried for help on translating parts of it.
1 3
and into two unequal numbers ad,db. But the additional three lines
are quite unusual. They are defined as follows:
e represents the square on gb
zh is the plane number from ad,db
hq is the square on gd
kl is the square on bd
nj is the square on dg
lm, mn are each the plane number obtained from bd,dg
According to the diagram (but not mentioned anywhere in the text)
kj = kl + lm + mn + nk. Notice then, that by virtue of II.4 and by
referring to the meaning attributed to each of the four segments,
Barlaam implicitly takes kj to represent the square on bd+dg. This is
nothing but the square on bg, which by definition is e. Also according
to the diagram (and not mentioned anywhere in the text) zq = zh +hq.
Thus, it is clear that the aim of the proof must be to show that also
zq = e. The steps of the argument for reaching this conclusion can be
schematically rendered as follows:
(o.1) By definition, kl = bd bd and lm = gd bd. Hence km = gb bd
[since km = kl +lm and gb = gd +db]
(o.2) But gb = ga, hence km = ga bd
(o.3) But by definition also, mn =gd bd. Hence by (o.1-o.2),
kn =ad bd
[since kn = kl +mn and ad = ag +dg]
(o.4) But also zh = bd ad. Hence [by (o.3)] zh = kn
(o.5) But hq = nj, since both were defined as cd cd. Hence kj = zq
[since kj = kn +nj and zq = zh +hq]
(o.6) But kj = e [by II.4 !!], so that [by (o.5)] zq = e
(o.7) But also zq = ad db +dg
2
[since in the diagram zq = zh +hq]
(o.8) Hence, zq = ad db+dg
2
and zq = e, so that ad db+dg
2
= gb
2
Barlaams proof presents another interesting example of an argument which is

certainly based on operating with numbers and relies on general properties of these
operations, but without thereby involving abstract manipulation of symbols. Nor does
there seem to be any kind of reliance on geometry. On the other hand, some of the
information necessary for completing the proof derives from the diagram alone and
is not explicitly mentioned in the text. Likewise, all the numbers involved are repre-
sented by straight lines, but some lines represent given numbers, while other represent
numbers that arise from operations on the given numbers. The two kinds are treated
differently. Worthy of special attention is the use of II.4 in step (o.6), since Barlaam
tacitly completes the step by referring to a number for which only the diagram indi-
cates the four factors that add up to it and that hence allow using the result previously
proved, II.4. Even the wording attached to the use of II.4 is different, since it states
that since the number bg has been divided into two numbers gd,db, therefore the
square on bg, namely e, is equal to the squares on bd,dg together with twice [the plane
number from] bd,dg.
Notice that, while the proof is stated in purely arithmetic terms and while some
of its steps amount to counting the units in each of the terms involved, it still seems
1 3
700 L. Corry
that in some crucial places the geometric background to the original propositions did
underlie the line of argumentation. Thus, like Gersonides, Barlaamproduced a version
of II.5 which was intended to assimilate this result into the body of arithmetic and
indeed, one might say, of Euclidean arithmetic. Their proofs, however, were different
as was also the fate of their works. Contrary to Gersonides book, Barlaams text
was well known in the sixteenth century to authors of inuential books, and some
of them, such as Petrus Ramus (15151573) or Christopher Clavius (15381612),
followed his formulations and his approach when handling II.5. Likewise, Sir Henry
Billingsley (c. 15451606), who in 1570 published the rst full English version of the
Elements, explicitly relied on Barlaams text (Billingsley 1570, Folio 62). Explaining
the details of Barlaams inuence on these authors, however, is beyond the scope of
the present article. We can say, at any rate, that this rather remote text appears to have
been instrumental in furthering the possibility of looking at II.5 from perspectives that
departed from the original, purely geometric one that underlies Euclids conception.
6 Concluding remarks
The rst printed version of the Elements appeared in 1482 in Venice, based on Cam-
panus translation, and it is usually known as the Erhard Ratdolt edition. The rst
published translation into Latin from a Greek text of the Elements appeared in 1505,
also in Venice. The Greek text had its origins in a slightly modied edition of Theon
of Alexandria. By comparing this Greek text with Campanus versions, its sixteenth-
century translator into Latin, Bartolomeo Zamberti (c. 1474after 1539), excluded
sections that in his opinion had been added by Theon [Rose 1975, 51 (note 56)]. Zam-
berti was aninuential humanist whoplayeda somewhat paradoxical role inpromoting
certain conceptions about the Elements at the time. On the one hand, he supported a
view, commonly associated with Proclus, that stressed the marvelous nature and
unity of the Elements. According to this view, the Euclidean text could not possibly
be rearranged or improved by means of new demonstrations without serious damage
to its perfection. On the other hand, Zamberti was also instrumental in popularizing a
view according to which, the choice of denitions, postulates, and propositions in the
Elements were Euclids, whereas the demonstrations were Theons. This alternative
view opened the way to editions in which the Euclidean text was altered, summarized,
divided, or published without the proofs (Goulding 2010, 152159).
In 1516, Jacques Lefvre dtaples attempted to reconcile the numerous discrepan-
cies between existing translations of the Elements and came up with a unied edition
in which the Radtolds and the Zambertis text appeared side by side. Eventually, these
two traditions, one stemming fromArabic sources (via Campanus) and one fromGreek
sources (via Zamberti), merged in the last third of the sixteenth century. In 1572, the
edition of Federico Commandino (15091575), based on this new merging of tradi-
tions, appeared in print, marking an important milestone in the process of assimilation
of the Elements as it came to be considered and understood in Europe.
The Euclidean traditions within the world of the printed text, which were inaugu-
rated with the 1482 edition and consolidated with the Commandino edition of 1572,
differed in important senses from the medieval ones that I have been discussing along
1 3
this article. This is certainly the case for the question of the changing relationship
among geometric and arithmetic ideas. For one thing, a main starting point of these
traditions was the Campanus text, and the peculiar way in which Book II and the ideas
related to it had been handled there. For another, new and vigorous trends of symbolic
algebraic techniques began to attract increased attention and were incorporated into
the text of the Elements. These new algebraic trends tted in a relaxed manner into,
and offered a natural continuation of, the kind of generalized arithmetic that Cam-
panus, especially under the inuence of Jordanus, had already made appear as the
fundamental way to look at the results of Book II. Within the renaissance traditions
of the Elements, then, Book II came to be understood in ways that went well beyond
the purely geometric one that had informed Euclids original conception. A detailed
analysis of this topic is well beyond the scope of this article, and I intend to pursue it
at a future opportunity.
Acknowledgments I want to pay my debt of gratitude to several friends and colleagues who read previous
versions of this text, or parts of it, answered queries and sent me reading material, raised questions, and
suggested ideas and advanced critical views: Fabio Acerbi, Sonja Brentjes, Karine Chemla, Menso Folkerts,
JanHogendijk, Jens Hyrup, TonyLevi, AnthonyLoBello, MarkMoyon, JeffreyOaks, Sabine Rommevaux,
Ken Saito, Shai Simonson, and Roy Wagner. Special thanks I owe to my friends Michael Fried and Miki
Elazar for help and patience with translations of difcult passages from Latin and Greek, and to Veit Probst
in Heidelberg, without whose kind mediation I would have had a much harder time in gathering all the
material needed to complete this work. Last, but not least, hearty thanks go to Len Berggren, to whom I
submitted this article for publication. His very detailed reading and thoughtful editorial suggestions helped
me signicantly improve (and somehow shorten) the entire text.
References
Archibald, Raymond C. 1913. Euclids book on divisions of gures with a restoration based on the Woepckes
text and on the practica geometriae of Leonardo Pisano. Cambridge: The University Press.
Billingsley, Henry. 1570. The elements of geometrie of the most auncient philosopher Euclide of Megara.
London. Imprinted by Iohn Daye.
Brentjes, Sonja. 1994. Textzeugen und Hypothesen zum arabischen Euklid in der berlieferung von al-
H
.
a g g a g b. Y usuf b. Mat
.
ar (zwischen 786 und 833). Archive for History of Exact Sciences 47: 5392.
Brentjes, Sonja. 1996. The relevance of non-primary sources for the recovery of the primary transmission
of Euclids Elements into Arabic. In Tradition, transmission, transformation: Proceedings of two con-
ferences on pre-modern science held at the university of Oklahoma, ed. F. Jamil Ragep, Sally P. Ragep,
and Steve Livesey, 201225. Brill: Leiden.
Brentjes, Sonja. 1997/1998. Additions to Book I in the Arabic traditions of Euclids Elements. Studies in
History of Medicine and Science 15(12): 55117.
Brentjes, Sonja. 2001a. Observations on Hermann of Carinthias version of the Elements and its relation to
the Arabic transmission. Science in Context 14(12): 3984.
Brentjes, Sonja. 2001b. Two comments on Euclids Elements? On the relation between the Arabic text
attributed to al-Nayrz and the Latin text ascribed to Anaritius. Centaurus 43: 1755.
Busard, Hubertus L.L. 1968. The translation of the Elements of Euclid from the Arabic into Latin by
Hermann of Carinthia (?). Edited with an Introduction. Leiden: Brill.
Busard, Hubertus L.L. 1983. The rst Latin translation of Euclids Elements, commonly ascribed to
Adelard of Bath. Toronto: Pontical Institute of Medieval Studies.
Busard, Hubertus L.L. 1984. The Latin translation of the Arabic version of Euclids Elements, commonly
ascribed to Gerard of Cremona. Leiden: Brill.
Busard, Hubertus L.L. 1987. The medieval Latin translation of Euclids Elements made directly from the
Greek. Stuttgart: Steiner Verlag.
1 3
702 L. Corry
Busard, Hubertus L.L. 1991. Jordanus de Nemore, De elementis arithmetice artis: A medieval treatise on
number theory, 2 vols. Stuttgart: Franz Steiner Verlag.
Busard, Hubertus L.L. 1992. The Arithmetica of Jordanus Nemorarius. In Amphora: Festschrift for Hans
Wussing on the occasion of his 65th birthday, ed. S.S. Demidov, et al., 121132. Basel-Boston-Berlin:
Birkhuser Verlag.
Busard, Hubertus L.L. 1996a. A thirteenth-century adaptation of Robert of Chesters version of Euclids
Elements. Munchen: Institut fur Geschichte der Naturwissenschaften.
Busard, Hubertus L.L. 1996b. Einiges ber die Handschrift Leiden 399,1 und die arabisch-lateinische
bersetzung von Gerhard von Cremona. In History of mathematics: States of the art, ed. J. Dauben, M.
Folkerts, and H. Wussing, 173205. San Diego: Academic Press.
Busard, Hubertus L.L. 2005. Campanus of Novara and Euclids Elements, 2 vols. Stuttgart: Franz Steiner
Verlag.
Busard, Hubertus L.L., and Menso Folkerts. 1992. Robert of Chesters (?) redaction of Euclids Elements,
the so-called Adelard II Version, 2 vols. Basel-Boston-Berlin: Birkhuser Verlag.
Clagett, Marshall. 1978. Archimedes in the Middle Ages. Volume iii: The fate of the medieval Archimedes,
13001565. Philadelphia: American Philosophical Society.
Corry, Leo. 2004. Modern algebra and the rise of mathematical structures, 2nd ed. Basel-Boston-Berlin:
Birkhuser.
Corry, Leo. forthcoming. The law of distributivity in the medieval traditions of Euclids Elements: Between
geometry and arithmetic. (Cited throughout as [LC2]).
Corry, Leo, and Norbert Schappacher. 2010. Zionist internationalism through Number Theory: Edmund
Landau at the opening of the Hebrew University in 1925. Science in Context 23(4): 427471.
Curtze, Maximilian. 1899. Anaritii in decem libros primos elementorum Euclidis commentarii ex interpre-
tatione Gherardi Cremonensis in cdice Cracoviense 569 servata, Leipzig: Teubner (Euclidis Opera
Omnia, I.L. Heiberg et H. Menge (eds.), supplementum).
Curtze, Maximilian. 1902. Der Liber Embadorum des Abraham bar Chijja Savasorda in der bersetzung
des Plato von Tivoli. Abhandlungen zur Geschichte der mathematischen Wissenschaften 12: 1183.
Djebbar, Ahmed. 2005. Lalgbre arabe. Gense dun art. Paris: Vuibert.
Dold-Samplonius, Yvonne. 1987. Developments in the solution to the Equation cx
2
+ bx = a from al-
Khwarizmi to Fibonacci. In From deferent to equant: A volume of studies in the history of science in
the ancient and medieval Near East in Honor of E.S. Kennedy, vol. 500, ed. David A. King, and Georg
Saliba, 7187. New York: Annals of the New York Academy of Sciences.
Eastwood, Bruce S. 2007. Ordering the heavens: Roman astronomy and cosmology in the Carolingian
renaissance. Brill: Leiden.
Folkerts, Menso. 1970. Boethius Geometrie II: ein mathematisches Lehrbuch des Mittelalters. Wies-
baden: Franz Steiner.
Folkerts, Menso. 1981. The importance of the Pseudo-Boethian Geometria during the Middle Ages. In
Boethius and the Liberal Arts: A collection of essays, Berne, ed. M. Masi, 187209, (Reprinted in Menso
Folkerts 2003), Essays on early medieval mathematics: The Latin tradition. Aldershot: Ashgate).
Folkerts, Menso. 1989. Euclid in medieval Europe. Winnipeg: The Benjamin Catalogue.
Folkerts, Menso. 2004. Leonardo Fibonaccis knowledge of Euclids Elements and of other mathematical
texts. Bollettino di Storia della Scienze Matematiche 24: 93114.
Folkerts, Menso. 2006. The development of mathematics in medieval Europe: The Arabs, Euclid, Regiomon-
tanus. Aldershot: Ashgate Publishing Limited.
Freudenthal, Hans. 1977. What is Algebra and What has it been in History. Archive for History of Exact
Sciences 16(3): 189200.
Fried, Michael N., and Sabetai Unguru. 2001. Apollonius of Pergas Conica: Text, Context, Subtext. Leiden:
Brill.
Goldstein, Bernard. 1965. A treatise on Number Theory from a tenth century Arabic source. Centaurus 10:
124160.
Goulding, Robert. 2010. Defending Hypatia: Ramus, Savile, and the renaissance rediscovery of mathemat-
ical history. Dordrecht, Heidelberg, London, New York: Springer.
Guillaumin, Jean-Yves. 2012. Boethius De institutione arithmetica and its inuence on posterity. In A
companion to Boethius in the Middle Ages, ed. Noel H. Kaylor, and Philip E. Phillips, 135162. Leiden:
Brill.
Guttmann, Yehiel Michael HaCohen. 19121913. Chibbur Ha-Meschicha WeHa-Tischboreth: Lehrbuch
der geometrie des Abraham bar Chija, vol. 2. Berlin: Schriften des Vereins Mekize Nirdamim.
1 3
Heath, Thomas. 1956 [1908], The thirteen Books of Euclids Elements. New York: Dover.
Heath, Thomas. (1981) [1921]. History of Greek mathematics, vol. 2. New York: Dover.
Heiberg, Johan Ludvig, and Heinrich Menge. 18831893. Euclidis opera omnia. Ediderunt et latine inter-
pretati sunt I. L. Heiberg et H. Menge, vol. 12. Lepzig: Teubner.
Hogendijk, Jan P. 2005. Al-Nayrzs inuence in the West. (Downloaded February 14, 2010, from http://
www.jphogendijk.nl/publ/naywest.pdf).
Hyrup, Jens. 1986. Al-Khwrizm, Ibn Turk, and the Liber Mensurationum: On the origins of Islamic
Algebra. Erdem 2: 445484.
Hyrup, Jens. 1988. Jordanus de Nemore, 13th century mathematical innovator. Archive for History of
Exact Science 38: 307363.
Hyrup, Jens. 1998. A new art in ancient clothes. Itineraries chosen between scholasticism and baroque in
order to make Algebra appear legitimate, and their impact on the substance of the discipline. Physis,
Nuova Serie 35: 1150.
Hyrup, Jens. 2001. On a collection of geometrical riddles and their role in the shaping of four to six
Algebras. Science in Context 14(12): 85131.
Hyrup, Jens. 2010. Hesitating progress: the slow development toward algebraic symbolization in abbacus
and related manuscripts, c. 1300 to c. 1550. In Philosophical aspects of symbolic reasoning in Early
Modern Mathematics (Studies in Logic, 26), ed. Albrecht Heeffer, and Maarten Van Dyck, 356. London:
College Publications.
Hughes, Barnabas. 1981. Jordanus de Nemore. De numeris datis. A Critical edition and translation.
BerkeleyLos AngelesLondon: University of California Press.
Hughes, Barnabas. 2008. Fibonaccis De practica geometrie. New York: Springer (Sources and Studies in
the History of Mathematics and Physical Sciences).
Kibre, Pearl. 1981. The Boethian De institutione arithmetica and the quadrivium in the thirteenth century
University Milieu at Paris. In Boethius and the liberal arts: A collection of essays, ed. Michael Masi,
6780. Bern/Frankfurt: Peter Lang.
Klein, Jacob. 1968. [19341936], Greek mathematical thought and the origin of Algebra (Translated by Eva
Brann. With an appendix containing Vietas Introduction to the analytical art, translated by J. Winfree
Smith). Cambridge, MA: MIT Press.
Lange, Gerson. 1909. Sefer Maassei Chosheb. Die Praxis des Rechners. Ein hebrisch-arithmetisches Werk
des Levi ben Gerschom aus dem Jahre 1321. Frankfurt am Main: Louis Golde.
Lefvre dtaples, Jacques. 1496. Arithmetica decem libris demonstrate. Paris: Johannes Higman and
Wolfgang Hopyl.
Levey, Martin. 1966. The Algebra of Ab u K amil. Kit ab f al-j abr wal-muq abala, in a commentary by
Mordecai Finzi (Hebrew Text, Translation and Commentary with Special Reference to the Arabic Text).
Madison: University of Wisconsin Press.
Lvy, Tony. 1992. Gersonide, commentateur dEuclide: Traduction annote de ses gloses sur les Elements.
In Studies on Gersonides: A Fourteenth-Century Jewish Philosopher-Scientist. Collection de Travaux
de LAcadmie Internationale dHistoire des Sciences, vol. 36, ed. Gad Freudenthal, 83148. Leiden:
Brill.
Lvy, Tony. 1997a. Les Elments dEuclide en hbreu (XIIIeXVIe sicles). In Perspectives arabes et
mdivales sur la tradition scientique et philosophique grecque, ed. M. Aouad, A. Elamrani-Jamal,
and A. Hasnaoui, 7994. Leuven: Peeters.
Lvy, Tony. 1997b. Une version hbraque indite des Elments dEuclide. In Les voies de la science
grecque, ed. D. Jacquart, 181239. Geneva: Droz.
Lvy, Tony. 1997c. The establishment of the mathematical bookshelf of the medieval Hebrew scholar:
Translations and translators. Science in Context 10(03): 431451.
Lvy, Tony. 2001. Les dbuts de la littrature mathmatique hbraque: la gomtrie dAbraham bar H
.
iyya
(XIeXIIe s.). Micrologus 9: 3564.
Lo Bello, Anthony. 2003a. Gerard of Cremonas translation of the commentary of Al-Nayrz on Book I
of Euclids Elements of geometry. With an introductory account of the twenty-two early extant Arabic
manuscripts of the Elements. Leiden: Brill.
Lo Bello, Anthony. 2003b. The commentary of Albertus magnus on Book I of Euclids Elements of geometry.
Leiden: Brill.
Lo Bello, Anthony. 2009. The commentary of al-Nayrz on Books II-IV of Euclids Elements of geometry.
With a translation of that portion of Book I missing from MS Leiden Or. 399.1, but present in the newly
discovered Qom manuscript ed. Rdiger Arnzen. Leiden: Brill.
1 3
704 L. Corry
Luckey, Paul. 1941. T abit b. Qurra ber den geometrischen Richtigkeitsnachweis der Ausung der
quadratischen Gleichungen. Schsischen Akademie der Wissenschaften zu Leipzig. Mathematisch-
physische Klasse. Berichte 93: 93114.
Mills-Vallicrosa, Jos Mara. 1931. Abraambar Hiia, Llibre de geometra. Hibbur hameixih uehatixboret.
Barcelona: Editorial Alpha.
Moyon, Mark. 2007. La tradition algbrique arabe du trait dAl-Khwrizm au Moyen ge latin et la place
de la Gomtrie. In Histoire et enseignement des mathmatiques. Rigueurs, erreurs, raisonnements, ed.
Evelyne Barbin, and Dominique Bnard, 289318. Lyon: INRP.
Mueller, Ian. 1981. Philosophy of mathematics and deductive structure in Euclids Elements. Cambridge,
MA: MIT Press.
Murdoch, John. 1966. Euclides Graeco-Latinus: A Hitherto unknown medieval Latin translation of the
Elements made directly from the Greek. Harvard Studies in Classical Philology 71: 249302.
Murdoch, John. 1968. The medieval Euclid: Salient aspects of the translations of the Elements by Adelard
of Bath and Campanus of Novara. Revue de synthse (ser.Il) l: 6494.
Murdoch, John. 1971. Euclid: Transmission of the Elements. Dictionary of Scientic Biography 4:
437459.
Murdoch, John. 2003. Transmission into use: The evidence of Marginalia in the medieval Euclides Latinus.
Revue dhistoire des sciences 56(2): 369382.
Neal, Katherine. 2002. From discrete to continuous: The broadening of the number concepts in Early
Modern England. Dordrecht: Kluwer.
Netz, Reviel. 1999. The shapingof deductioninGreek mathematics: Astudy incognitive history. Cambridge:
Cambridge University Press.
Oaks, Jeffrey. 2001. Geometry and proof in Ab u K amils Algebra. In Actes du 10me Colloque Maghrbim
sur lHistoire des Mathmatiques Arabes, 234256. Tunis: LAssociation Tunisienne des Sciences Math-
matiques.
Oaks, Jeffrey. 2011. Al-Khayy ams scientic revision of Algebra. Suhayl: Journal for the History of the
Exact and Natural Sciences in Islamic Civilisation 10: 4775.
Puig, Luis. 1994. El De Numeris Datis de Jordanus Nemorarius como sistema matemtico de signos.
Mathesis 10: 4794.
Rabinovitch, Nahum L. 1970. Rabbi Levi ben Gershon and the origins of mathematical induction. Archive
for History of Exact Sciences 6: 237248.
Rommevaux, Sabine. 1999. La proportionnalit numrique dans le Livre VII des lments de Campanus.
Revue dhistoire des mathmatiques 5: 83126.
Rommevaux, Sabine. 2008. The transmission of the Elements to the Latin West: three case studies. In
The Oxford handbook of the history of mathematics, ed. Eleanor Robson and Jackie Stedall, 687706.
Oxford: Oxford University Press.
Rommevaux, Sabine, Ahmed Djebbar, and Bernard Vitrac. 2001. Remarques sur lHistoire du Texte des
Elments dEuclide. Archive for History of Exact Sciences 55: 221295.
Rose, Paul Lawrence. 1975. The Italian Renaissance of mathematics: Studies on humanists and mathemati-
cians from Petrarch to Galileo. Genve: Librairie Droz.
Rosen, Frederic 1831. The Algebra of Mohammed ben Musa. London: Murray.
Saito, Ken. (2004) [1985]. Book II of Euclids Elements in the light of the theory of conic sections. His-
toria Scientiarum 28: 3160 (Reprinted in Jean Christianidis (ed.), Classics in the History of Greek
Mathematics, Boston Studies in the Philosophy of Science, Vol. 240, Dordrecht-Boston: Kluwer,
pp. 139168).
Saito, Ken, and Nathan Sidoli. 2012. Diagrams and arguments in ancient Greek mathematics: Lessons
drawn from comparisons of the manuscript diagrams with those of modern critical editions. In The
history of mathematical proof in ancient traditions, ed. Karine Chemla, 135162. Cambridge: Cambridge
University Press.
Sarfatti, Gad B. 1968. Mathematical terminology in Hebrew Scientic Literature of the Middle Ages
. Jerusalem: Magnes Press.
Sesiano, Jacques. 1987. Survivance mdivale en Hispanie dun problme n en Msopotamie. Centaurus
30(1): 1861.
Sesiano, Jacques. 1993. La version latine mdivale de lAlgbre dAb Kmil. In Vestigia Mathematica.
Studies in Medieval and Early Modern Mathematics in Honour of H. L. L. Busard, ed. Menso Folkerts,
and Jan P. Hogendijk, 315452. Amsterdam et Atlanta: Rodopi.
1 3
Sigler, Laurence E. 2002. Fibonaccis Liber Abaci. A translation into Modern English of Leonardo Pisanos
Book of Calculation. New York: Springer (Sources and Studies in the History of Mathematics and
Physical Sciences).
Simi, Annalisa. 2004. Leredit della Practica geometriae di Leonardo Pisano nella geometria del basso
Medioevo e del primo Rinascimento. Bollettino di Storia della Scienze Matematiche 24: 943.
Simonson, Shai. 2000a. The missing problems of GersonidesA critical edition. Historia Mathematica
27(3): 243302.
Simonson, Shai. 2000b. The missing problems of GersonidesA critical edition. Historia Mathematica
27(4): 384431.
Tummers, Paul M.J.E. 1980. The commentary of Albert on Euclids Elements of Geometry. In Albertus
Magnus and the Sciences: Commemorative Essays, ed. J.A. Weisheipl, 479498. Toronto: Pontical
Institute of Mediaeval Studies.
Tummers, Paul M.J.E. 1994. Anaritius commentary on Euclid. The Latin translation IIV
(Artistarium Supplementa IX). Nijmegen: Ingenium Publishers.
Unguru, Sabetai. 1975. On the need to rewrite the history of Greek mathematics. Archive for History of
Exact Sciences 15: 67114.
Unguru, Sabetai and David Rowe. 19811982a. Does the Quadratic Equation have Greek Roots? Libertas
Mathematica 1: 149.
Unguru, Sabetai, and David Rowe. 19811982b. Does the Quadratic Equation have Greek Roots? Libertas
Mathematica 2: 162.
van der Waerden, Bartel L. 1976. Defense of a Shocking point of view. Archive for History of Exact
Sciences 15: 199210.
Vitrac, Bernard. 2003. Les scholies grecques aux lments dEuclide. Revue dhistoire des sciences 56:
275292.
Vitrac, Bernard. 2004. A propos des dmonstrations alternatives et autres substitutions de preuves dans les
lments dEuclide. Archive for History of Exact Sciences 59: 144.
Vitrac, Bernard. 2011. Faut-il rhabiliter Hron dAlexandrie? Montpellier: France. (Downloaded May 02,
2012, from http://hal-paris1.archives-ouvertes.fr/hal-00454027).
Vitrac, Bernard, and Maurice Caveing. 1990. Euclide dAlexandrie. Les lments. Traduit du texte de
Heiberg (Vol. I. Introduction gnrale par Maurice Caveing. Livres IIV: gomtrie plane. Traduction
et commentaires par Bernard Vitrac). Paris: Presses Universitaires de France.
Vlasschaert, Anne-Marie. 2012. Le Liber Mahameleth: dition critique et commentaries. Stuttgart: Franz
Steiner Verlag.
Weil, Andr. 1978. Who betrayed Euclid? Archive for History of Exact Sciences 19: 9194.
Weinberg, Josef. 1935. Die Algebra des Ab u K amil o g aben Aslam. Mnchen: Druck der Salesianischen
Ofzin.
1 3

Medieval Arithmetic Geometry

Uploaded by

Copyright:

Available Formats

Medieval Arithmetic Geometry

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Medieval Arithmetic Geometry

Uploaded by

Copyright:

Available Formats

Arch. Hist. Exact Sci.

II.10: If a straight line is bisected, and a straight line is added to it in a straight

The second argument relies on the equality of the rectangles RA and

In the proof of A-I.20 Jordanus invokes some additional results such

.2) a c = e(by def.), but a c = d b; hence d b = e and d = b +e

.3) a d = f(by def.), but a d = c b; hence c b = f and c = b +f

Barlaams proof presents another interesting example of an argument which is

You might also like