Journal of Philosophy, Inc.

Russell's Theory of Types

Author(s): Edwin Guthrie
Source: The Journal of Philosophy, Psychology and Scientific Methods, Vol. 12, No. 14 (Jul. 8,
1915), pp. 381-385
Published by: Journal of Philosophy, Inc.
Stable URL: http://www.jstor.org/stable/2013671 .
Accessed: 24/05/2014 13:00

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal
of Philosophy, Psychology and Scientific Methods.

http://www.jstor.org

This content downloaded from 91.229.248.39 on Sat, 24 May 2014 13:00:44 PM

All use subject to JSTOR Terms and Conditions
 PSYCHOLOGY AND SCIENTIFIC METHODS 381

are. No uniformity may ever be unity; yet, all the more, in this
world of many different uniformities there is unity and there is the
necessity that unity enjoins and that all uniformities serve or medi-
ate. Long live this "Power behind the Throne"; behind all thrones;
a power whose only law, the only absolute law, is the principle of
law and order or, just once more, the "synthetic unity a priori."
Finally, thus to be able to translate the old-time Kantian Trans-
cendentalism into the recent creative evolution, to be able to read
in Kant's causality as category a priori the nature of causation as
involving (1) duality, (2) incommensurability, and (3) necessity,
but vital necessity, is at once not without a large tribute to the
"vision," if possibly not the clear seeing, of Kant and decidedly
with a most significant historical justification of the new creationism
and its great retinue of other important "isms."
ALFRED H. LLOYD.
UNIVERSITY OF MICHIGAN.

RUSSELL'S THEORY OF TYPES

M R. RUSSELL 'S solution of the paradoxes collected in the
"Principles of Mathematics," through his theory of types,
has been received with little comment other than explanatory notices,
which is a situation that is somewhat puzzling, for the consequences
of the theory for logic extend far beyond the few paradoxes for which
it was invented, and its detailed statement makes an elaborate intro-
duction to the algebra of logic that it would be pleasant to avoid.
The problem of the paradoxes is the old problem of the insolubilia
that formed the text for many chapters in scholastic logic, and the
principle of the vicious circle, that no function may have itself as an
argument, calls to mind Peter von Ailly's "Pars propositionis non
potest supponere pro toto," and the provision that a "function is not
determinate unless its values are previously determinate,"' recalls
the scholastic doctrine of restriction which made a verb in the present
tense apply only to the instant before its utterance, not the time of
utterance, and so did away with the paradox of the liar. In the
theory of types we have Peter's maxim worked into a consistent and
thorough device that successfully treats the paradoxes. A typical
instance of the paradoxes is, "This proposition is false," which if
false can be proved true, and if true can be proved false. If we rep-
resent "this proposition" by P and "false" by f we may write it:
(P<f) < (P<f)' and (P<f)'< (P<f). One of these expres-
sions would have given no paradox, for that a proposition implies its
contradictory means only that it is false; but that its contradictory
1 "v Principia Mathematica, " page 43.

This content downloaded from 91.229.248.39 on Sat, 24 May 2014 13:00:44 PM

All use subject to JSTOR Terms and Conditions
382 THE JOURNAL OF PHILOSOPHY

also implies the proposition results disastrously. Let r be such a

proposition; we then have r=r', and on multiplying by r, r 0, or
on multiplying by r', r'= 0, and so r + r'- 0; but by the law of
contraction r + r'_ 1. Hence we have 1 0, which is a manifest
paradox. Russell finds the root of the paradoxes in their self-ref-
erence. This proposition could have been derived from the proposi-
tional function, "There is a proposition x and x is false," and the
paradox results from allowing the function itself to be a value of x.
This possibility must be avoided, and is avoided by the theory of types
which establishes certain ranges of significance for the variables in all
functions or propositions. If the function or proposition itself is al-
lowed to be the valiue of one of its terms a paradox results. We shall
then make this impossible by the principle of the vicious circle, " What-
ever involves all of a collection must not be one of the collection. "
Thus the range of significant values of a variable in a function must
exclude the function itself or anything derived from the function.
This exclusion is accomplished in detail by the hierarchy of types,
whose exact statement would be too long to repeat here. The general
outlines of the theory are as follows: All individuals (entities that
are not propositions or functions) constitute a group, the first log-
ical type; all propositions about individuals only, or functions that
have as possible values for their terms individuals only, are first-order
propositions or functions, and themselves constitute the second logical
type. Propositions or functions which take as values for their terms
not only individuals, but first-order functions or propositions, consti-
tute second-order propositions or functions, a higher logical type.
To be predicative or legitimate a proposition can have as values of its
terms only members of a type lower than itself. Only such proposi-
tions are meaningful, or true or false. "This proposition is false,"
being of order n and having the same proposition of order n as one
of its terms is meaningless. "I lie" becomes "I am making a state-
ment of order n and it is false," which is false merely, for no state-
ment of order n is made: the actual statement being about an nth-
order proposition is itself of order n + 1 and not n.
The interest of this, theory for general logic which deals with
propositions only-for propositions alone can be true or false-lies
in the fact that a proposition in order to be meaningful must allow
terms of a lower order only. This is sufficient to solve the paradoxes,
but sufficient to do much more than that. We have "solved" a large
number of statements whose need of solution was decidedly question-
able. Mr. Shearman, in his "The Scope of Formal Logic," has
called attention to the fact (which Russell has also remarked) that
there are propositions thus made illegitimate that lead to no paradox,
and cites the "law of contradiction" that all propositions are true or

This content downloaded from 91.229.248.39 on Sat, 24 May 2014 13:00:44 PM

All use subject to JSTOR Terms and Conditions
 PSYCHOLOGY AND SCIENTIFIC METHODS 383

false. This proposition is to be outlawed by the theory of types, since

its reference to "all propositions" includes itself and so violates the
principle of the vicious circle. Mr. Shearman suggests that the
theory of types needs limitation, but he does not offer any criterion
that will select the paradoxical propositions for elimination and leave
those that are harmless.
Any reference to "all propositions" has been declared meaning-
less. More than that, since a proposition about "some propositions"
is equivalent to the denial of a statement about "all propositions,"
that also is meaningless. In fact, any statement about "proposi-
tions," becomes impossible. "There exists a proposition" is equiva-
lent to "It is false that all propositions are non-existent. " Since the
latter is illegitimate, the first is also. Moreover, in the statement,
" ' "X implies Y" is a proposition' is a proposition," the two terms
written "proposition" are in different types, in different universes,
and it is difficult to see how they can have any common meaning or
can be compared. They were better represented by different words.
A like situation holds with respect to truth.2 "It is evident," says
Russell, "that the definition of trut7t is different in the case of gen-
eral judgments from what it was in the case of elementary judg-
ments. Let us call the meaning of truth which we gave for elemen-
tary judgments 'elementary truth.' Then when we assert that it is
true that all men are mortal, we shall mean that all judgments of the
form 'x is mortal,' where x is a man, have elementary truth. We
may define this as 'truth of the second order' or 'second-order
truth.' " Judgments about truths of the second order will consti-
tute third-order truth. An assertion of the truth of a statement can
not have truth in the same sense in which it uses the term. There
seems to be nothing common to first-order truth and second-order
truth except that they are designated by the same name which the
theory leaves an accident or a mystery. The difference in type leaves
no community whatever.3
Self-reference of this sort may be at the root of the paradoxes, but
it exists in many propositions without apparently making them
meaningless. " This proposition is true" does not yield any paradox;
it may be held meaningless because it has no reference outside itself,
which is John Wyelif's method of dealing with the paradoxes in his
"Tractatus de Logica"; but in the statement "This proposition has
five words" we seem to detect something that could be meant. The
2 "' Principia Mathematica, " page 48.
3 The effect of the theory on Cantor's theory of transfinite ordinals is equally
astonishing. The first transfinite ordinal (w) being the ordinal type of all
finite ordinals is in a different universe from these, and its addition to a finite
ordinal or multiplication by a finite ordinal (which are necessary to the deriva-
tion of new transfinites) becomes meaningless.

This content downloaded from 91.229.248.39 on Sat, 24 May 2014 13:00:44 PM

All use subject to JSTOR Terms and Conditions
384 THE JOURNAL OF PHILOSOPHY

proposition refers to its own expression. Apparently the theory of

types would make it meaningless. One more trivial example: would
not the statement, "This article is too long by so much as this sen-
tence," appended to an article inathis JOURNAL be true and derive
its significance from its self-reference?
Perhaps it would be a mean spirit that would ask whether the
prineiple of the vicious circle, that no proposition can refer to itself,
refers to itself in speaking of "no proposition," or whether, since
each implication in logic contains as unexpressed premises all the
axioms and principles that apply to the case, Russell has not made
each proposition say in effect: " This does not apply to itself. " These
questions can be overlooked in the in-terest of a solution. But there
are more serious difficulties. "If mathematics is to be possible, it is
absolutely necessary ... that we should have some method of making
statements which will usually be equivalent to what we have in mind
when we (inaccurately) speak of "all properties of x." (A state-
ment about all properties of x is itself a property of x, and so is il-
legitimate under the theory of types.) This is to be remedied by the
axiom of reducibility. This assumes "that every propositional func-
tion is equivalent, for all its values, to some predicative function of
the same arguments. " 4 Stated symbolically:

H:(3f)
:cox.o,-f!x.
The question occurs: If only predicative propositions are to be true
or false, how can we have a non-predicative proposition formally
equivalent to a predicative proposition, that is, true when it is true,
false when it is false? That would seem to require a true non-
predicative proposition, which is a contradiction. The principle of
the vicious circle requires that non-predicative functions be meaning-
less. Yet reference is made repeatedly to terms which satisfy a non-
predicative function. This occurs in the definition of identity and else-
where :5"x and y are to be called identical when every predicative func-
tion satisfied by x is also satisfied by y. We can not state that every
function satisfied by x is also satisfied by y, because x satisfies fune-
tions of various orders, and these can not all be covered by one ap-
parent variable. By virtue of the axiom of reducibility it follows
that, if x = y and x satisfies fx, where / is any function, predicative
or non-predicative, then y also satisfies 'y. Hence, in effect, the defi-
nition is as powerful as it would be if it could be extended to cover
all functions of x. " What satisfaction of a non-predicative function
can mean is not clear. If it means "makes the function a true prop-
osition" we may have self-reference in a true proposition resulting
4 " Principia Mathematica, " page 174.
'5 "Principia Mathematica," page 176.

This content downloaded from 91.229.248.39 on Sat, 24 May 2014 13:00:44 PM

All use subject to JSTOR Terms and Conditions
 PSYCHOLOGY AND SCIENTIFIC METHODS 385

from a non-predicative function. If it means "makes the function a

true proposition without paradox" the whole question is begged.
The axiom of reducibility seems to be an acknowledgment that the
theory of types is too " drastic" that does not offer any effective relief.
EDWIN GUTHRIE.
UNIVERSITY OP WASHINGTON.

REVIEWS AND ABSTRACTS OF LI'TERATURE

Der Gegenwartswerthder geschlichtlichen Erforschung der mittelalter-
lichen Philosophie. M. GRABMANN. Akademische Antrittsvorlesung.
Wien: Herder. 1913. Pp. 94.
Comparativeestimates in philosophyare receivedusually with caution;
for the task is difficult, and the provincial temper is persistent. But an
attempt to find in the philosophy of another time,-and that the Middle
Ages,-direct value for the solution of present-dayproblems,is, very liable
to be ignored as presumptuous. In reality, however, this brochure is
adapted to perform an important service; and the circle of its appeal is
wider than might appear. Its spirit and form are of the kind permissible
after researchin a given field is well under way, and before the results are
finally in. Accordingly, its value lies in the stimulus furnished, rather
than in the broadclaims made. In this sense, the able mustering of facts
and reflections will appeal to all who are interested in ideas and their
history; and the suggestions concerningthe continuity of thought and the
value of the past will be welcomed. The workmay serve, also, for orienta-
tion in this field, as a supplementto the larger worlisof Picavet.1 In turn,
its interpretativeaspect may be supplementedby the works of Farge, of
Sentroul, and of P. Rousselot, for example.
The writer is an authority in his own field,2 who is also au courant
with contemporarythought. And his work affords,on the whole, a happy
illustration of the neo-scholastic'scapacity for the objectively historical
attitude. The nature of the present work makes it partly polemical; but
the polemic is directed to adherent (e. g., pp. 41 ff.; cf. p. 92) as well as
contemner of the " philosophia perennis." He urges the "stilling of all
prejudiceon both sides, with an eye single to faithful pictures of scholastic
and of contemporarythought" (p. 46; cf. pp. 32, 44), and a mutual under-
standing in the commonsearch for truth (p. 78; cf. p. 94). The work is,
therefore, significant as part of a rapprochement;the catholic scholar is
losing his fear of the new, and his non-catholic colleague is regaining his
faith in the old. The shadow of the Renaissance is passing. And with
objective historicity on both sides established, we may hope to see our
1 " Esquisse, " etc., 2 ed., 1907, and " Essais, " etc., 1913.
2 Grabmann 's most important work is the "Gesch. d. schol. Methode," Bd.
I., 1909, Bd. II., 1911, Bd. III., promised. The next in importance is his "D.
philos. u. theolog. Erkenntnissth. d. Kard. Math. v. Aquasparta," 1906. Other
works: on Thomas Aquinas, Ulrich of Strassburg, John the German, and the mod-
ern scholastic activity.

This content downloaded from 91.229.248.39 on Sat, 24 May 2014 13:00:44 PM

All use subject to JSTOR Terms and Conditions