Early  Journal  Content  on  JSTOR,  Free  to  Anyone  in  the  World  
This  article  is  one  of  nearly  500,000  scholarly  works  digitized  and  made  freely  available  to  everyone  in  
the  world  by  JSTOR.    

Known  as  the  Early  Journal  Content,  this  set  of  works  include  research  articles,  news,  letters,  and  other  
writings  published  in  more  than  200  of  the  oldest  leading  academic  journals.  The  works  date  from  the  
mid-­‐seventeenth  to  the  early  twentieth  centuries.    

 We  encourage  people  to  read  and  share  the  Early  Journal  Content  openly  and  to  tell  others  that  this  
resource  exists.    People  may  post  this  content  online  or  redistribute  in  any  way  for  non-­‐commercial  
purposes.  

Read  more  about  Early  Journal  Content  at  http://about.jstor.org/participate-­‐jstor/individuals/early-­‐

journal-­‐content.    

JSTOR  is  a  digital  library  of  academic  journals,  books,  and  primary  source  objects.  JSTOR  helps  people  
discover,  use,  and  build  upon  a  wide  range  of  content  through  a  powerful  research  and  teaching  
platform,  and  preserves  this  content  for  future  generations.  JSTOR  is  part  of  ITHAKA,  a  not-­‐for-­‐profit  
organization  that  also  includes  Ithaka  S+R  and  Portico.  For  more  information  about  JSTOR,  please  
contact  [email protected].  
 On the Surfaces with Plane or Spherical Curves of
Curvatfure.
BY PROF. CAYLEY.

The theory is considered in two nearly cotemporaneous papers-Bonnet,

"Memoire sur les surfaces dont les lignes de courbure sont planes ou spheriques,"
Jour. de l'Ecole Polyt. t. XX (1853), pp. 117-306, and Serret, " Memoire sur les
surfaces dont toutes les lignes de courbure sont planes ou spheriques,"Liouville,
t. XVIII (1853), pp. 113-162. J desire to reproduce in a more compact form,
and with some additional developments, the chief results obtained in these
elaborate memoirs.
The basis of the theory is a theorem by Lancret, 1806. In any curve
described upon a surface, the angle between the osculating planes at consecutive
points is equal to the difference of the angles between the. osculating planes and
the correspondingtangent planes of the surface.
This includes as a particular case Joachimsthal's theorem, Crelle,t. XXX
(1846): If a surface have a plane curve of curvature, then at any point thereof
the angle between the plane of the curve and the tangent plane of the surface
has a constant value.
Bonnet and Serret each deduce the like theorem for a spherical curve of
curvature, viz: If a surface have a spherical curve of curvature, then at any
point thereof the angle betweenithe tangenitplane of the sphere and the tangent
plaineof the surface has a constant value. Bonnet (Me6moire,p. 235) says that
this follows from Lancret's theorem. Serret (Memoire,p. 128) obtains it, by the
transformationby reciprocal radius vectors, from Joachimsthal's theorem.
I remark that the theoremnfor a spherical curve of curvature, and (as a par-
ticular case thereof) that for a plane curve of curvature, are obtained at. once
from the most elementary geometrical considerations, viz: if we have (in the
same plane or in different planes) the two isosceles triangles NPP', OPP' on a
commlonbase PP', then the angle OPN is equal to the angle OP'N. For take
72 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature.

P, P' consecutive points on a spherical curve of curvature; then at P, P' the

normals of the surface meet in a point N, and the normals (or radii) of the
sphere meet in the centre 0, and we have angle OPN= angle OP'N, that is, at
each of these points the inclination of the normal of the surface to the normal of
the sphere has the same value; and this value being thus the same for any two
consecutive points, must be the same for all points of the curve of curvature.
The proof applies to the plane curve of curvature; but in this case the
fundamnentaltheorem may be taken to be, a line at right angles to the base PP'
of the isosceles triangle NPP' is equally inclined to the two equal sides
NP, NP'.
A surface may have one set of its curves of curvature plane or spherical.
To include the two cases in a comnmnon formula, the equation may be written
k(x2 +y2 + z2)- 2ax- 2by- 2cz- 2-= 0; lc=1 in the caseof asphere,= 0
in that of a plane; and the expression a sphere may be understoodto include a
plane. I write in general A, B, C to denote the cosines of the inclinations of
the normal of the surface at the point (x, y, z) to the axes of coordinates (con-
sequently A2 + B2 + C2 1). Hence considering a surface, and writing down
the equations
k (x2+ y2 2ax - 2by- 2cz-
+ Z2) - = 0,
(kx -a) A + (7ky-b) B + (kz -c) C= 1,

where (a, b, c, u, 1)are regarded as functions of a parameter t, the first of these

equations is that of a variable sphere; and the second equation expresses that at
a point of intersection of the surface with the sphere, the inclination of the
tangent plane of the surface to the tangent plane of the sphere has a constant
value 1, viz: this is a value depending only on the parameter t, and therefore
constant for all points of the curve of intersection of the sphere and surface: by
what precedes, the curve of intersection is a curve of curvature of the surface,
and the surface will thus have a set of spherical curves of curvature.
Supposing the surface defined by means of expressions of its coordinates
(x, y, z) as functions of two variable parameters, we may for one of these take
the parameter t which enters into the equation of the sphere; and if the other
parameter be called 0, then the expressions of the coordinates are of the
form x, y, z = x (t, 0), y(t, 0), z (t, 0) respectively; these give equations
dx, dy, dz-adt+a'dO, bdt+b'dO, cdt+c'dO (where of course (a, b, c, a', b', c')
are in general functions of t, 0), and we have A, B, C7 proportional to
 CAYLEY: On the Surfaces witthPlane or Spherical Curvesof Curvature. 73

be'- b'c, ca' - cta, ab'- atb, viz: the values are equal to these expressions each
divided by the square root of the sum of their squares. In order that the
surface may have a set of spherical curves of curvature, the above three equa-
tions must be satisfied identically by mealnsof the values of
a, b, c, u, 1, A, B, C, x, y, z

as functions of (t, 0); and it may be seen without difficulty that we are thereby
led to a partial differential equation of the first order for the determination of
the surface. But I do not at present further consider this question of the deter-
mination- of a surface having one set of its curves of curvature (plane or)
spherical.
Suppose now that there is a second set of (plane or) spherical curves of
curvature. We have in like mianner
x(x + y2 + z2) - 2ax-2y-2yz-2v ?0,
(xx -) A + (xy - 3) B + (xz - y') C- X =O,

where x is 1 or = 0 according as the curves are spherical or plane, and

(a, d37', v, A) are functions of a variable parameter 0. We take the t of the
formnerset of equations and the 0 of these equations as the two parameters in
terms of which the coordinates (x, y, z) are expressed. This being so (the
former equations being satisfied as before), if these equations are satisfied identi-
cally by the values of a, y I', v, Aj,A, B, C, x, y, z as functions of (t, 0), then
the surface will have its other set of curves of curvature also spherical. It will
be recollected that by hypothesis a, b, c, u, I are functions of the parameter t
only, and that a, d3,y, v, X functions of the parameter 0 only. The foregoing
equations, together with the assumed relations
A2+B2+ C2=1,
Acdx + Bdy + adz! 0,

are the "six equations" for the determination of a surface having its two sets of
curves of curvature each of them (plane or) spherical.
Assuming now the values of a, b, c, 1, u as functions of t, and a, 7
/,7 v
as functions of 0, the question at once arises whether we can then satisfy the
six equations. These equations other than Adx + Bdy + Cdz-- 0, or say the
five equations, in effect determine any five of the eight quantities A, B, C,
10
74 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature.

x, y, z, t, 0, in terms of the remaining three, say they determine A, B, C, t, 0 as

functions of x, y, z: we thus have a differential equation Adx + Bdy + Cdz = 0
wherein A, B, C are to be regarded as given functions of (x, y, z). An equa-
tion of this form is not in general integrable; and if the equation in question
be not integrable, then clearly the system of equations cannot be satisfied by
any value of z as a function of (x, y), or, what is the same thing, by any values
of (x, y, z) as functions of (t, 0). We thus arrive at the condition, the equation
must be integrable, viz: the condition is
(dB dC\ du dA\ (dA- dBN _

V A z y-d) + B dx d C d d
If this be satisfied, then we have an integral equation 1= 0 (containing a con-
stant of integration which is an absolute constant) and which is in fact the
equation of the required surface. But it is proper to look at the question
somewhat differently. Supposing that the condition V = 0 is satisfied, then we
have the integral equation 1=0 O, and this equation, together with the five equa-
tions, in effect determine any six of the quantities A, B, C, x , y, z, t, 0 in
terms of the remaining two of them, or, what is the same thing, they determine
a relation between any three of these quantities. We can, from the five equa-
tions and their differentials, and from the equation Adx + Bdy + Cdz= 0,
obtain a differential equation between any three of the eight quantities: and it
has just been seen that corresponding hereto we have an integral relation
between the same three quantities; that is (the condition V= 0 being satisfied),
we can from the six equations obtain between any three of the quantities
A, B, C7,x y, z, t, 0 a linear differential equation of the foregoing form (for
instance Zdz + Tdt+ ?dO= 0, where Z, T, 0Eare given functions of z, t, 0) which
will ipSofacto be integrable, furnishing between z, t, 0 an integral equation which
may be used instead of.the before-mentioned integral equation 1= 0. And we
thus have (without any further integration) in all six equations which serve to
determine any six of the quantities A, B, C7, x, y, z, t, 0 in terms of the
remairlingtwo. It is often convenient to seek in this way for the expressions
of (A, B, C and) x,y, z as functions of t, 0 in`preference to seeking for the
integral equation 1= 0 between the coordinates x, y, z.
The condition V = 0 is in fact the condition which expresses that at any
point of the surface the two curves of curvature intersect at right angles.
Serret (and after him Bonnet) in effect obtain the condition by the assumptior
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 75

of this geometrical relation, without showing that the geometrical relation is in

fact the necessary condition for the coexistence of the six equations. They give
the condition in the form dAxx+ dySy + dzz = 0, where dx, dy, dz are the
increments of (x, y, z) along one of the curves of curvature, and Sx, Sy, Sz the
increments along the other curve of curvature. The equations give

(kx - a) dx + (ky - b) dy + (kz- c) dz = 0,

Adx + Bdy + Cdz - O,
and similarly
(xx - a)Ax + (xy - 13) y + (xz - ') &z= 0,
Ax + By + Sz_ O.
We thence have
dx: dy dz =
B (kz - c)-C (ky-b): C(kx -a)- A (kz- c): A (ky - b)-B (kx-a),
and
& y: -

B (xz-y) -
C(xy-3): C (xx-a) -A (xz -y): A (zy /3)
-
B(xx -a).

We have thus the required condition, in a form which is readily changed into
(A2+B2+ C2) (kx-a)(x -a) +(ky-b)(xy-3) + (kz- c)(xz -7)
-{A (kx-a)+B(ky- b) +C(kz-c)} I A(xx-a) +B (xy-3) + C(xz-7) J= 0,

and writing herein A2+ B2 + C2= 1, this becowes

1
x {k (x2 + y2 + z2) - 2ax - 2by- 2czf

+ 4 k {X (X2 + y2 +z2)- 2ax- 2y- 2yz}

+ (aa + bfi + cy) - 2 =0,
that is,
aa + b6 + cy - la+ c + kv- 0.

I proceed to show that this is the condition V = 0 for the integrability of the
differential equation Adx + Bdy + Cdz= 0. Writing as before
/ dB dC\ / d dA) dA dB
dz dy / dx dy dx
76 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature.

we have from the six equations

AdA + BdB + CdC= o,
(kx - a) dA + (ky - b)dB + (kz - c) dC
- k (Adx + Bdy + Cdz) + (Aal + Bb1 + Cc, + 11)dt,
(xx-c ) dA + (xy - f) dB + (xz-r) dC
-- (Adx + Bdy +Cdz) + (Aa' + Bgt3+C(1 +J) dO,
(kx- a) dx +(ky- b)dy +(kz- c)dz=(alx+bly+c1z+u1)dt,
(xx - ) dx + (xy - f dy + (xz -ry) dz = (o'x + f'y + )'z + v) dO,
where a,, bl, cl 1,, ul denote derived functions in regard to t and a/, p3',y, 2/, v'
derived functions in regard to 0. Putting for shortness
A, B, C,
n- kx -a, ky-b, kz-c
XX a, xy-g, XZ-)y
we readily obtain
= [(xy- j) C- (xz- 7)B]
?LdA {-k (Adx+ Bdy+ Cdz)
+ (al+ Bbk+ c, + 111(7cx-a) dx + (ky-b) dy + (kz-c) dz} }

-[(ky-b) C-(7kz-c) B]{- x (Adx+ Bdy + Cdz)

a/+X+ jj++,' {xx--a) dx + (xy-g)dy+ (xz-)

- dz5) };
say this is
fIdA = [(xy- () C - (xz - ') B] {- k (Adx + Bdy + Cdz)
L
+ p {(kx-a)dx+(ky-b)dy+(kz-c)dz}

-[(ky-b) C-(kzz-c) B] -x (Adx + Bdy + Cdz)

+ A 1 (xx-a)dx+ (xy-/3)dy+ (xz-y)dz4t

or introducing further abbreviations, and writing -down the analogousvalues of

fQdB and Q2dC,we have
fQdA= [(xy - ) C - (xz- -7) B] U- [(ky-b) C -(kz-c) B] T,
VdIB= [(xz- -)A- (xx-a)C] U- [(kz - c) A- (7x -a)ClrT,
QdC = [(xz - a) B - (xy-3) A] U [(k-x- a) B -(ky-b) A] T.
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 7 7

We hence find
L
dz = [(xz-y)A- (xx-a) C] kC + (kz-c))

_[(kz-c)A-(kxG-a)C] -xC+ AJ (xz-y)

-Qi E=-(xx -a) B-(xy-, A] - k-B+ -p- (ky -b))

+ [(kx-a)B-(ky-b)A] (-xB + A
(xy_)
(dB dC\
and combining these two terms, ini the resulting value of a ( d - first
the term without L or A is found to be
-kA A (xx- a) + B (xy- )+ C(xz-y)
-k (xx- a)(A2+ B2+ C2)
+ x (kx- a)(A2?+ B2+ C2)
+ xA [A (7kx- a) + B (ky-1b) + C(7z -c) ,
which is
- -kA + k (xx - a)- x (kx - a) + xAl,
A(xl- ka) - ka + xa.

Next the coefficient ofL is

A (kz - c)(xz - y) - C(xx - a)(kz - c)

+ A (ky -b)(xy- f) -B (cx a) (ky-b),
which is
= A [(kx - a)(xx - a) + (ky - b)(xy -f3) + (kz - c)(xz - y)]
(xx- a)[A (kx - a) + B(ky -b) + C(kz-c)]
=AM+ (xx-c a) l,
if for shortness
M= (kx - a)(xx - a) + (ky - b)(xx -) + (kz - c)(xz -y);

and similarly the coefficient of Ais

- A(kz -c)(xz -y) + C(kx -a)(xx- a)

-A(ky - b)(xy -f3)+ B(kx -b)(xy -)
78 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature.

which is
- A [(kx - a)(zx - a) + (ley- b)(xy -) + (kz - c)(xz )]
(kx - a)[A (xx- a) + B (xy - + C(xZ - )]
- AM- (kx -a) X.
We thus obtain

QK dz -dy) =A (xl-kX)-kca + xa +L JAM+ (xx-a) 1l

A
IAM+(kx -a)2X},
and similarly
/dB dAL
(dz- dz ) - (xl-k%)-kl+ xb + {BM+ (xy-fl) 1l
r { BM+ (ky -b)X ,
L
f( dA dB Cz
dx (xl-kX)-ky+xc +
p
ICM+(xz- /)l
-y
A
-
I- {CM+ (kz c) X,
and hence multiplying by A, B, C and adding, we obtain
QV-xl- k- k (Aa + Bf + Cy) + x (Aa + Bb + Cc)

+ L (M- 1) - A (M_ X),

where the first four terms are together
Xz1-k k+ kc x(Ax + By + Cz)-t - x{k(Ax + By + Cz) 1$,
viz. these destroy each other, and the equation becomes
fIV Lp- II)(M 1%).
But we have
M- %= 1 xick(x; + y' + z2) - 2ax -2by -2ezt
+ + k {x (x2+ y2 + Z2) 2ax - 2,y - 22yzt+ (aa + b,6+ cy) -
-

which is -aa + b + cy-l1 + xi + kv,

or we find L A (aa + + cy-1, + zu + kv)

viz. the condition V =0 is

aa + 6j +cy -l++xu +kv 0,
the result which was to be proved.
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 79

If we consider separately the cases where the two sets of curves of curva-
ture are each plane, the first plane and the second spherical, and each of them
spherical; or say the cases PP, PS and SS, then in these cases respectively the
condition is aa + bg3+ cy - l72= 0,
aa + b, + cy 1-t + u = 0,
aa + bg + cy - 1i + u + V =0:
we have in each case to take the italic letters functions of t and the greek
letters functions of 0, satisfying identically the appropriate equation, but other-
wise arbitrary; and then in each case the six equations lead to a differential
equation Adx + Bdy + Cdz= 0 (or say Zdz + Tdt + @4O= 0) between three
variables, which equation is ipsofacto integrable; and we thus obtain a new inte-
gral equation which with the original five integral equations gives the solution
of the problem. The condition is in each case of the form 2aa = O, the number
of terms aa being 4, 5 or 6. Considering for instance the form
aa + b3+cy+dA+ ee+fcp=O
with 6 terms, it is easy to see how such an equation is to be satisfied by values
of a , b c, d , e, f which are functions of t, and values of a, d3,y (3, E, (p which
are functions of 0. Suppose that t1, t2, .... are particular values of t, and a1,
b, X.... .I; a2 , b2
62 *- - Al etc., the corresponding values of a, b, .... f, these
values being of course absolute constants; we have a, .... cp, functions of 0,
satisfying all the equations
(a,, bl, C17 dl, ej,fiXa, kt,y,Us, 6,El) o,
(a2, b2, C2, d,2, e2,f2 ,
etc.,
and if 6 or more of these equations were independent, the equations could, it is
clear, be satisfied only by the values a = y = (= E= (p= 0. To obtain a
proper solution, only some number less than 6 of these equations can be inde-
pendent. Suppose for instance only two of the equations are independent; we
then have a, 93, y , E, (pfunctions of 0 satisfying these equations, but otherwise
arbitrary; or, what is the same thing, we may take a,j3, y, 3, E, (plinear func-
tiolns of 6 - 2, 4 arbitrary functions, say P, Q, R, S of 0; say we have
a
.(o.
a3.P. a, ,2, Q, R, S),
a
(o07 ..... > " ),
* * * * * * * * * * * * * *

* * * * * * * * * * * * * *

=
*p ((Po **** **.
80 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature.

where the suffixed greek letters denote absolute constants; and this being so, in
order to satisfy the proposed equation aa + b,3 + cy + df3+ ee + fp =0, we
must have
yo 7o, cox cpa, b, c, d, e, f)-
(aox09 7o 0,
(., ....... .. ")
(a2,S )0 -,....3 s
(a3 .,

viz. a, b, c, d, e, f will then be functions of t satisfying these four equations, but

otherwise arbitrary. The above is a solution for the partition 2 + 4 of the
number 6. We have in like manner a solution for any other partition of 6; or
if we disregard the extreme cases a = b = c = d = ef =f O and a = G =y =S
= = = 0, then we have in this manner solutions for the several partitions
15, 24, 33, 42 and 51 of the number 6.
But applying this theory to the actual problem, there is a good deal of
difficulty as regards the enumerationof the really distinct cases. I use the letters
P, S to denote that a set of curves of curvature is plane or spherical as the case
may be, the surfaces to be considered are thus PP, PS, and SS. First, for the
PP problem where the equation is aa + b3 + cy - = 0, the two systems
(a , b c, 1) and (a, iB,y, A) are symmetrically related to each other, and instead of
the solutions 13, 22 and 31, it is sufficient to consider the solutions 13 and 22.
But here (a , b c, 1) are not a system of four symmretricallyrelated functions,
(a, b, c) are a symmetrical system, and I is a distinct term: and the like for the
systenm(a, j3, y, A). In the PS problem, where the equation is aax+ b63+ cy
-1i + u = O, and thus the systems (a, b, c, 1, u), (a, 3, y, A, 1) are of
different forms, we should consider the solutions 14, 23, 32 and 41: but here
again in each of the systems separately the terms are not symmetrically related
to each other. Lastly, in the SS problem where the equation is aa + b/ + cy
- ia + u + v=O, the systems (a, b, c, 1, u, 1) and (a, &,y/, A, 1, v) are of the
same form, it is enough to consider the solutions 15, 24 and 33; but in this case
also in each of the systems separately the terms are not symmetrically related to
each other. I do not at present further consider the question, but simply adopt
Serret's enunmeration.
It is to be remarked that for a developable (but not for a skew surface) the
generating lines may be curves of curvature, and regarding the generating
lines as plane curves we might have developables PP or PS; but a straight
 (JAYLEY: On the Surfaces with Plane or Spherical Curves of Curvature. 81

line is not a curve in a determinateplane, and it is better to consider the case

apart from the general theory. Again, the curves of curvature of one set or
those of each set may be circles; anida circle may be regarded either as a plane
or a spherical curve; regarding it, however, as a spherical curve, it is a curve not
in a determinatesphere. The cases in question, of the curves of curvature of the
one set or of those of each set being circles, are therefore also to be considered
apart from the general theory. The surfaces referred to present themselves for
consideration among Serret's cases PP 10, 20, 30; PS 10, 20, 30, 40, 50, 60, 70;
and 88 10, 20, 30, 40; but they are excluded from his enumeratioin, and he in fact
reckons in his " Conclusion,"pp. 161, 162, two kinds of surfaces PP, three kinds
PS, and two kinds SS.
It is very easily seen that if a surface has a plane or spherical curve of
curvature, then on any parallel surface the corresponding curve is a plane or
spherical curve of curvature: and thus if a surface be PP, PS, or SS, then the
parallel surfaces are respectively PP, PS or SS. The solutions obtained
include for the most part all the parallel surfaces, and thus there is no occasion
to make use of this theorem; but see in the continuation of the present paper
the case considered under the subheadingpost, PS40 = Serret's third case of PS.
If a surface have a plane or spherical curve of curvature, then transforming
the surface by reciprocal radius vectors (or inverting in regard to an arbitrary
point), then in the transformed surface the corresponding curve is a spherical
curve of curvature. Hence if a surface be PP, PS or SS, the transformed
surface is 8S. Conversely, as shown by Bonnet and Serret, and as will appear,
every surface SS is in fact an inversion of a surface PP or P8.
I proceed to the enumeration, developing the theory only in regard to the
two, three, and two, cases PP, PS and SS respectively.

PP, THE Two SETS OF CURVES OF CURVATURE EACH PLANE.

The six equations are

A2+B2 +C2 -1,
ax + by + cz + u 0,
Aa+Bb +Cc+ =O,
acx + f3y + yz + v-0,
Aa +B13 +Cy+ O,
Adx + Bdy + Cdz =0,
11
82 CAYLEY: On the Surfaces with Plane or Spherical Curves of Curvature.

the condition is
aa + b61+ cy - 1= 0
(not containing u or v, so that these remain arbitrary functions of t, 0 respec-
tively). The cases are
a b c 1 a a
PP1? 1 0 c 0 0 1 0 a
PP2? 0 1 0 -m a -ma y a
PP3? 1 0 c mc 0 1 ma a
m is an arbitrary constant, and in the body of the table c is an arbitrary function
of t, and a, y/, a arbitrary functions of 0.
PP1? is Serret's first case of PP, included in his second case.
PP20 gives developable.
PP3? is Serret's second case of PP.

I consider the case

PP30 = SERRET'SSECONDCASE OF PP.

1
Writing for greater symmetry m = g,-= =f, so that fg = 1; also ma = y,
and consequently a = fy , we take c and y for the two paramietersrespectively,
or write c = t, 7 0; also changing the letters u , v, we write
a b c 1 u a 3 y v
-1, 0, t gt P 0 1 0 fO II

and the six equations thus are

A2+B2 +C2 =1,
x+tz -P =0,
A + tC -gt = 0,
y+ Oz - =- ,
B+0C f0 =0,
Adx + Bdy + Cdz= O.

We seek for the differential equation in z, t, 0. We have

A2 + B2 + C2 = 1, A = t(g -C), B= 0(f -C),
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 83

and thence t2 (g- C )2 + 02(f -C) + C2_ 1i

that is C2 (1 + t2 + 02) + 2C(gt2 + f02) = 1 - g2t2-f202,
or multiplying by 1 + t2+ 02and completing the square,
(1 + t2 + 02) C - gt2-f022- (1 g2t2- f202)(1 + t2 + 02) (gt2 + f02)2 -

= Xf+ (f -g) t23lg + (g + (gfA) 02)

1
if~~~~ 1

= f +(f g) t2'
1 _ g + (g _A 02

and thence giving a determinate sign to the square root, say

(1 + t2+ 02) Cgt2+f-+ f

an equation which may also be written
C fT- -gE
f-g
In fact, observing that T2 - 02 (f1- g)(1 + t' + 02),we deduce from the
original form

(2 - 2) a =
(f g)(t2+ f02) jj

or thri n ot te
or f2 g

(T E))(T +{)
or tbrowing out the factor T + ,E- and reducing, we have the required value;
and thence forming the values of A anidB, we have

A =-tT f e , B = - OE) f
T - 9' T--e E' ,' a =fT-_ gO.@
T-?E'
we have, moreover,
x+tz-P, y+z=H11,
or differentiating,and writing P1 and I' for the derived functions in regard to
t and 0 respectively,
dx =-td,z-zdt + Pldtt, dy- 'Odz -zd,0-H dO.
 84 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature.

The equation Adx. + Bdy + Cdz= 0 thus becomes

-
-Ttdx - EOdy+ fT gE)dz = 0
f- g
viz. this is
[- t T(- tdz - zdt + Pldt) -06 (- Odz- zd0 + H'dt)] (f- g) + (fT- gO) dz = 0,
or collecting,
[(f + (f- g) t2) T- (g + (g -f)) 0] dz + (tTzdt + O6zdO)(f - g)
(tTP1dt+ 06H'dO)(f-g) 0,
that is,
/1 1\
(S- T ) dz + (f- g) z (t Tdt + 06d0) - (f - g)) (t TPdt + 00171'd0)= 0,
which is an integrable form as it should be, viz. the equation is
d - I) z - (f- g)(tTP1dt + 0611'dO)= 0
and we obtain
(T- )z - (f g)f (tTPidt + 061dO) = 0
the constant of integration being considered as included in the integral. But
it is proper to alter the form of the second term. Take F, cFarbitrary func-
tions of t, 0 respectively; and writing F1, b' for the derived functions, assume
p , rI =- ; we have

f(tTP1dt + OErl'dO)
= (gtT(T ) dt + f00 Q() do)
f0cF
F + gtFi, D-( + <,2
In fact this will be true if only

(-F + T) = gtT T3 (- + .'2 )=f06 E)

which are equations of like form in t, 0 respectively, and it will be sufficient to
verify the first of them. Effecting the differentiation, the terms in Fl1 destroy
each other, and there remain only terms containing the factor F1, and throwing
this out we obtain
1+ T2 + T3
viz. this is
-1 +g(f+(f -g)t2) -gt2(f g)= 0,
which is identically true, and the equation is thus verified.
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 85

The foregoing result is

( 1
(D\) z+(f-g) F1t+cl gtF _0,

we then have
x=+ tz=g0 X Y+ Z- =0 ,

and hence, repeating also the equation for z,

(4 - 1)x+(f-g){-t(-F+c) gtF + }+( -1 +=o,

(T (9) +(f g){ - F+(D)gtol -3 } 0,

equations which give the values of the coordinates x, y, z in terms of the

parameters t, 0. It will be recollected thatfg = 1 (f or g being arbitrary), that
the values of T, ? are

t
T2 f+ (f - g)t2, s g + (g f) 0,
and that F, 'D denote arbitrary functions of t, 0 respectively. I repeat also
the foregoinigequations
-
A, B, C=- tTf_9
T -e'
-
-f} OEfT- 0'3 fTT- e(3(i
The equations may be presented under a different form; we have

-tTx- -OE)y+ fT_ g z + F+ D=o,

-fT3(x + tz) +F 0,
-ge3 (y + Oz) + '= 0,

where it will be observed that the second and third equations are the derivatives
of the first equation in regard to t and 0 respectively. We thus have the required
surface as the envelope of the plane represented by the first equation, regarding
therein t, 0 as variable parameters. Moreover,the second equation (which contains
only the parameter t) represents the planes of the curves of curvatureof the one
set; and the third equation (which contains only the parameter 0) represents the
86 CAYLEY: On the Suifaces with Plane or Spherical Curvesof Curvature.

planes of the curves of curvature of the other set. It is to be observed that from
the equations for 1, X (viz. A + tC=gt and B + OC=f0) it appears that for
any plane of the first set the inclination to a tangent plane of the surface is
=cos- gt and that for any plane of the second set the inclination is

CO- Vf+d
=cos' z1 + 02.

It may be remarked that the last mentioned results may be arrived at by

the consideration of an equation Ax + By + Cz + D = 0, where the coefficients
are functions of t and 0 (A a function of t only, and B a function of 0 only) such
that the derived equations A1x + GC1z +D1 = 0 and B'x + C'z +D' = 0 depend
the former of them upon t only, and the latter of them upon 0 only.
A very simple case of the equation is when f= g = 1; here T= 0 = 1,
and the surface is the envelope of the plane z - tx - Oy+ F + FD = 0.
Returning to the general form

-tTx -OE)y + fT-g) z + F+ 4) =

I transform this by introducing therein in place of t, 0 two variable parameters

a, (3 which are such that ka =-tT, kg3= 0 (k a constant which is presently

put = 1f )' we find

= fk2a 2 02i gk2'3
1-(f-g) T 2?' (g-f) k32 '
and thence
1 -1
T Vi1- (f g)k2o- (9- V1 (gf )
k22,
4/f 4/g

or puttingk 1 these last values are

1 1
T-~ ,y^~ -2~ ~ F ~~A /1 + MI,
and we hence obtain
fT-gE 3 IN/f a?-/ V 1 +(2,
1- N/
f-gq =f-g f
- 4/f~g{4f~g4/1 a2 4/ V+ g2,
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 87

say this is k.I%/F1-a2 -yV1 +[2},

where X = =f It __g , and therefore 2 _ 2 1 or y =VA2 _ 1.

Vf -g' 1Vf -y

Hence writing F+?D = k (A + B), k times the suin of two arbitrary func-
tions of a and d3respectively, the equation becomes
ax dy + z t%s/i 71 + 2l + A + B=-
viz. the surface is given as the envelope of this plane considering a, d3 as two
variable parameters. This is the solution given by Darboux, "Le,cons sur la
theorie generale des surfaces, etc.," Paris, 1887, pp. 128-131. He obtains it in a
very elegant manner, starting fromnthe followinig theoremn:Take A, Al, etc.,
functions of the parameter a, and B, B1, etc., functions of the parameter g3;then
if we have identically
(A1 - B1)2+ (A2 - B2)2 + (A3- B3)2 -(A4 - B4)2,

the required surface will be obtained as the envelope of the plane

(A1 - B1)x + (A2-B2)y + (A3 -B3)z = A - B,

where A, B are two new functions of a, / respectively.

The foregoing identity is the condition in order that each sphere of the
one series (x - A1)2+ (y - A2)2+ (z - A3)2 -= A2 nay touch each sphere of the
other series (x - B1)2+ (y - B2)2+ (z - B3)2= B2; the two series of spheres
thus envelope one and the same surface which will have its curves of curvature
of each set circles: viz. this will be the surface of the fourth order called
Dupin's Cyclide, the normals whereof pass through an ellipse and hyperbola
which are focal curves one of the other, and which contain the centres of all the
spheres touching the surface along its curves of curvature. The equations of
the ellipse and hyperbola may be taken to be

(X2 + Z2 =1 y = 0) and (y2 1, x=0)

respectively, and we thence obtain the required PP surface as the envelope of

the plane
aux- gy + (V/1 a2 - /V-2 -i1VI + 32)z + A + B 0.
88 CAYLEY: On the Surfaces with Plane or Spherical Cu,rvesof Curvature.

THE CASE PP10 = SERRET'S FIRST CASE OF PP.

We deduce this from the second case by writing therein m = 0, that is,
g = 0, f = oo; but it is necessary to make also a transformation upon the
parameter 0, viz. in place thereof we introduce the new parameter p, where
02 02~ gep
tgc2' this gives

2
192= f + (g _ f) 02= g (( + (g (l 2_
-_ gcP2

and thence

=
06 0=Vg +(g
+( f) 02
fT0=77
VfV1iq2~ -2 ff -g gE) for g 0 is = T.
We have also T= Vf+ (f g)t2X = VfVi + t2 when g = 0, and substi-

tuting these values, considering (F as a function of cp, and for F + CF writing

as we may do F + (, the equation becomes

-t my + z F+cF0
^/f^/V
+ tf %If l
,/ t2 % Vf
where the divisor VIf is to be omitted. Hence finally, instead of p restoring the
original letter 0, and again considering 'D as a ftinction of 0, the equation is
z- tx Oy
F+ D
Vi+t2~~~~0
s/1 t2 4,/ 02+ =o,
viz. here F, 1 are arbitrary functions of t, 0 respectively, and the surface is the
envelope of this plane considering t, 0 as variable.
We obtain an imaginary special form of PP10 by writing in this equation
kOfor 0 and then putting k = o; the ci reinains an arbitrary function of the new
0, and the equation is
tx
- ~ ~~ ^/1 Y F+

( =s/ - 1 as usual). This is in fact the equation which is obtained from PP30
by simply writing therein g = 0 without the transformationupon 0.
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 89

PS. THE SETS OF CURVES OF CURVATURE, THE FIRST PLANE, THE SECOND
SPHERICAL.
The six equations are
A2+B2+ C2 -1,

ax+ by + cz + u o0,
Aa +Bb+ Cc + - 0,
x2 + y2 + Z2- 2ax - 23y - 27z - 2v = 0,
A(x-a) +B(y {()+ C(z-y)-2 =0,
Adx + Bdy + Cdz -0.
The condition is
aa + b3 + cy- 1 + u== 0,

(not containing v so that this remains an arbitrary function of 0). The cases are

a b c I u i oc r
PSi, a b c 0 0 0 0 0
PS2? a b c I ml 0 0 0 m
PS30 (a b c -mc 0 0 0 y y
PS40 a b 0 1 ml 0 0 y m
PS50 a b 0 0 0 0 0 y
PS60 0 b 0 1 ml a 0 y m
1
Ps70 a b 0 ma 0 a 0 y - a

where m is an arbitrary constant and in the body of the table the other italic
letters are arbitraryfunctions of t, and the greek letters arbitrary functions of 0.
PSi0 is Serret's first case of PS, included in his second case.
P82? gives developable.
PS30 is Serret's second case of PS.
PS40 is Serret's third case of PS.
PS50 gives circular sections (surfaces of resolution).
PS60 gives circular sections (tubular surfaces).
PS70 gives circular sections.

12
90 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature.

I consider
PS3?= SERRET'SSECONDCASE OF PS.

The six equations are

A2 +B2 + C2 = 1
ax + by + cz =0,
Aa+Bb + Cc- cm,
x2+Y2 +(z-mcp)2=-O+m 2,
Ax +By + C(z- mp) =p,
Acdx+Bdy + Cdz 0,
(where a, b, c are assumed such that a2 + b2+ c2 1). We easily obtain
(1 - c2) A = - ac (C + m) -b,
(i-c2)B - bc(C+m) +aV&f,
and thence
aB -bA = /V,
where
=(1 -c2)(1 - ) -c2(C+ m)2 c2: C_C2 2Cm
c -c2m2;
also
xV/ - c2in2 = ApVi - c2m2 + (bC cB) V0 + (2 1) (p2,

YVl - m
C22 BPV - cIm2 + (cA - aC) VO + (M2 - 1) 2,

z~Vi-c2m2- (C+ m)qp1 -c2n-2 + (aB -bA) VO +(m2 -)qp2.

We seek for the differential equation in C, t, 0. From the equation
Ax + By + (C- mp) z- ,
and attending to Adx + Bdy + Cdz= 0, we deduce
xdA + ydB + (z - mqp)dC- (1 + Cm)p'dO= 0,
and we have herein to substitute for dA, dB their values in terms of d , dt, do.
We have AdA + BdB - Cdc,
adA + bdB -cdC - Q,
if for shortness Q = Ada + Bdb + (Ca + i) dc. Hence
V dA = (-cB + bC)dC -BQ,
VndB (-= aC+ cA) dC + AQ.
We find without difficulty,
(1 -- c2) Q = (C + in) c/c+ (adb- bda)Vf?,
 CAYLEY: On the Surfaces with Plane or Spherical Cu-rvesof Curvatuare. 91

and consequently,
(1-c2)/ dA= b(C +c'm) -acV/f a4dC
-B (C + m) dc + (adb -bda)Vs/,
(1 -c') VQ dB -a (C + c2in) - bcV? } dC

Substituting these values we have + A I(C + m) dc + (adb - bda)

V/f
(bx - ay)(C + c2m) - (ax + by) cVJ tdC
- (Bx - Ay){(C + m)dc + (adb- bda)V?IQ}
+(1 c')V,/fI(z-mcp)d C-(1+ Cm)cp'dO =O,
viz. this is
(bx -ay)(C + c2m) - (ax + by)cV?iQ+ (1 c2)(z - mq)VSI C
-(Bx -Ay)I(C+m)dc+ (adb -bda)V/fl
-(1- c2)(1 + Cm)VQicp'dO 0.
The coefficient of d C contains a term - (ax + by + ez) cV,2 which is=0.
Moreover, we have
C + c2m
bx-ay =-_WZ+ 2 + (M2-1) V2
and then
(1-c2)(Bx -Ay) = c(C+ m)(bx ay) -czVD..
C + c2m
--c (C + m) { QV+ /0 + (m2-1)q2}
Vri- - c2m2

-cVD/4l(C+ m)qp + V 1/+c(2m )2j

which, observing that the terms in Ccp,/Vfdestroy each other, and that we have
(C+ m)(C+ c2m) + Q = (1 -c2)(1 + Cm), gives

- c(1 + CM) 0 +
c2m2VO+(n->p(q2
-)2
Bx-Ay=~I%V1
and the equation becoines

-(_ + %f ++ C2I/V + (M2 1)p2(C+ C2M) +zV?1

~(1 c2)pID i 2}dC

-co(1 + Cm) j>0+(m 1P2 {(C + m) dc + (adb- bda)V%l2}

-(1 - c2)V (1 + Cq) >p'dO O.
92 CAYLEY: On the Surfaces with Plane or Spherical Curves of Curvature.

Here the coefficientof dC is = [z-(C + m) p]vf + (C + cm) V/0 +(m2M2_ )p2

or substituting for z - (C + m)V/f its value = + I and

1 _ C2m2
V%
observing that Q + (C + c2m)2 = (1-c2m2)(1 C2), this coefficient is found to -

be = i _- c2n.2 (1 - c2) V/0 + (M2 _ 1>p2,

p and we have
V1 -c2m2-2 (1 2) VO + (m2 - 1) cp2dC

-c (l + Cm) ^/
0 + (m2 _1 p2 { (C + n) dc + (adb-bda)V/fi

(1- C2)(1 + GCm)Vcp'dO =0-,

or as this may be written
1/J1. -c2rn2dC (C + rn) cdc c (adh - bda)
/ tl 1 +Cm(1cmc2)V1 C2m2 (1 - c2)V1 C2M2
cp'dO
VO+ (m2_ 1)ip2= ?
where from the foregoing value of Ql we have identically
?2 (1 - I) = (1 _-C2)(1 + Cm)2 - (1 - c2m2)(C + M)2.

Here a, 6, c are functions of t and we have thus the required differentialequa-

tion in C, t, 0.
It is convenient to inultiply by the constant factor V 1- i2. The first
term is an exact differential, viz. writing
V1-C2m2 C+m M_ _ _n__ _
Cm' and thereforecos =
C2(I +V CM)'7
sin = / -
_c2 C21 +1 CM -
~Vi
we have
- M~n2VI
VI_s/1 6-2M2 dCa (C + gn)cdc
d ^-V ,7 l{V1 + Cmd + (1 c2)1 '-2M2
as may easily be verified. And the second and third terms are obviously the
differentials of a function of t and a function of 0 respectively. But to obtain
the integral functions, a transformationof each term is required.

7
First, for the term V/1 -m2 c (adb -6da)
~ ~ ~ (-C2)VA/I i we take a,b
C2rn2 ; etk
a b c functions
uctoso of t -

which are such a2+ b2 + c2= 1; and then writing a1, b1, c1 for the derived
functions so that aa1+ bb1+ cc1=0, we assume a', b', c' = Va1, Vb1, Vc, where
1
2 2
+ C2 we 2+ + c'2
V2- a + bi + c we have therefore aa' + bb' + cc' = 0, and a2+ b'2 '=1
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 93

and then writing a", b1, c"= bc'- b'c, ca'- c'a, ab' - a'b respectively, we
have aa" + bb" + cc"= 0, a'a"+ b'b"+ c'c"= 0, a"12+ b"I2+ c"2=1; thus
a, b, c, a', b', c', a", b", c" are a set of rectangular coefficients. We then write
a b, c- (a' + mb"), (b' - ma"), - c',
determining p so that a2+ b2+ c _ 1 as above, viz. we thus have
p2 (1 + cm), + c'2m2.
Observe that we thus have p2(1 - c2) = p2- c'2 and p2(1-c2m2) = (1 + cm)2.
Writing now
T= tan-1 c"ac+ nd therefore Sii, c+ cos T=c
, /
V i2'V p2- c'2 V9P2
we find that V/ 1-m2 c (adb - bda)
dT
(1- C2)V 1 c2m2
The verification is somewhat long, but it is very interesting. We have
l
dT- mi In2{c"dc - (c + m) dc"d
p2 c12

or observing that c" ab' - a'b, = V(abl - alb), dc = cldt, this is

Vi - .2
d T = p-2 2z V(ab - alb) el- (C + m)[ V, (ab - alb) + V (abl - allb)] dt,
where we have
= a2+ b2+ c2, and therefore-Vi. alaV,
= + + clell;
b1bIl
also frolm aa1+ bb1+ cc, = 0, we have a' + bl + cl + aall + bbl + cc,1=O,
and we thence obtain
_
dT- /Vi- ,2V7dt - (ab1- alb) cl (aa,, + bbl + cc,,)
p2 c12
- (c + m)[- (a1all + bib,, + 1lcl)(ab, - alb) + (a' + b' + c1)(abli - ball)] $,
the term in [] is found to be l- cI all(bel-blc) + bl, (cal- cla) + cln(abl- alb) },
hence cl appears as a factor of the whole expression, and reducing the part inde-
pendent of m, we find
dT= Vi1- ml1V3c1dt
p2 c12 (alb1l- allbl) + m [a1,,(bc, - b1c)
+ bl, (cal - c1a) + cil (ab, - a1b)] }.
Next calculating the value of adb - bda, we have
V V
a= {al+r(cal-c1a)I, b= jbi-m(bci-bjc)},
P p
94 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature.

or as these may be written

V V
a= p-a1(1 + cm)-ac,m}, b= - b1(I+cm)-bbemJ,

and we thence easily obtain

adb - bda= V2 (1 + cm)galbil - allbl)

+ m [a,, (bc1- b1c) + b1l(ca1- c1a)+ c1l(ab1- alb)] },
viz: the factor in } has
b the same value as in the expression for dT, and we
thus have dT 1-m2 Vcop2 IcV l-m2
adb -bda (1 + cm)(p2 c'2) Vi I c2m2 (1 c2) )

that is, dT=Vi - m2 c (adb-bda)

(1-c2)V1 c2Mn2
the required equation.
Secondly, for the term Vim (idO we introduce 1 a function of 0,
VO/ + (in2 - 1) qp2'
such that writing (V'for the derived function we have

?> = /1-4 (1-m2) qq! + 4 (1-M2) - ,/ suppose,

whence also
1) 4:2
43
1- 2O(D'
S/ 0 + (M2_ 4)p2= /0 + (M2 _

1%/M V/0+(M2-j)4p2-VO+(M2-j)C]2'

Then writing 1 m2 0 + (m2 _ 1)qj2

Vi 0- 41/ -
Vo

sin0 /
I , cos 0= O _
0 + (M/2

we find
=_ I
_1 -20c)dO
cos dE m2(?

that is, = I- in2 (1'- 20V) dO

@- 0v~O./ + (m2 --j) (D2

and similarly cos 60 dOTV =

-200')
0( 2 dO

that is, dO0 - V

V
-
m2(p - 204p')dd
0= O (M2_ 1p2
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 95

Hence
Vi -m2l r - 20_:F' p - 20p' d
clO+d00 20 lye/0+ (m2 _~ 1)PiI2 4/0 + (m2 - 1) 21

V - m2 (p -( -20q)) p1-m2 V p'dO

-:- 20 1 /0 + (m2_ 1)-02f - +-(1)m)
(A2 2'
the required equationl.
We find, moreover,
2DV/V1 qm2/0+ (m2- 1) 1- 2(1-m2)cIx
sin (E- 00)-=I/ M I,cs (E - 00) -M
which will be presently useful.
The differential equation now is d, - dT+ dO - dO = 0, hence the inte-
gral equation (taking the constant of integration = 0) is < T - 0 + 0o, or
say
sin '-sin (T- 8 + 00),
viz. substituting for sin T anldcos T their values, and observing that
/l-c2m2 C+m _ 1 +cm C+ m
sin< V
A1-c2 1 + cm' VP, c12 1+ Cm

the factor V29 multiplies out, and we have

(1 + cm) 1 + cm = (c + m) cos (0 - 00)-c"V/1-m2 sin (0-00).

And I further remark here that a former equation is

- m2) (1 c2)(1 + Cm)2-(1 - c2m2)(C+ rm)2,
thatQai(1
is,
- m (-c2m2)(C +m)1 (1 c2) cos24.

We thus have
1+Cm
+ cm Vp2 c

p4/+i - m2 {C"Vi- m2 cos (Q-QO) + (c + m) sin (Q-O- }.

We have thus C, and consequently also A, B, x, y, z all of them given as
functions of t, 0; but the formulae admit of further development.
Write C+ m ____-_
WriteC + m' whence also C -m
96 CAYLEY: On the Surfaces with Plane or Spherical Curves of CPurvature.

We have C(1-mn) + m =t, and hence (1+ cm) C (1 -me) + m , = (1 + cm) 8,

=(c + m) cos () -00) - c"/1 -m2 sin (E -0). Using the value of C
given by this equation, and calculating from it those of A, B; then writing for
shortness
X = a/ 1 -m2 cos (0-0) (a"-mb') sin (-0 -0),
Y= bVi/ - 2 cos ( -00) -(b" + ma') sin (- 00),
Z-= (c + m) cos (E --0) -c"V/ -m2 sin (E-00),
we have
A (1-m)(1 + cM)=V-m/ _ 2X,
B(1-ma)(1 + cM) = m2Y,
C(i mt)( + cm) = Z- m (1 + C)
to which I join U (1 + cm)-= Z.
By way of verification observe that Al + B2 + C2= 1, and that the equations
give
(1 _mt)2(1 + cm)2 (1 m2)(X2+ Y2+Z2) + M2Z2- 2mZ(1+ cm) + m2 (1 + Cm)2;
we have X2 + Y2 + Z2 = (1 + CM)2, Z= (1 + cm), and hence the identity
(1 - nw)2(1 + cm)2 (1 - m2 + m9 - 2nwa+ m2)(1+ cm)2.
Proceeding to calculate the values of x, y, z, recollecting that V/ 1 - c2m2
1
- (1 + cm), we have
x(1 +cm)-=A4(1 +cm) +p(bC-cB) V/O+ (m2 1)>p2,
-Aqp(i + cm) + {(b'- ma")C- c'B^VO0+ (M2I 1)q2,
that is,
Z(1 + cm)(l- ms) = (PS/ 1 M2X + 1 + Cr I b'-mall)(Z -m (+ cm))

-c'V-m2 y}'/ + (m2-1)i2

=pVi/ l-m2X + 1 4cm (b'-ma") Z cV 1-_M2 Y}/ 0 + (m2 _1)2

-m (b'- ma") V/O + (m2-1) p2,

where the term (b' - ma")Z - c'V 1 - m2 Y contains the factor 1 + cm; in fact
this is
=(b'- ma")1 (c + m) cos (E-030)- ,/1W-qm2 sin (0E- ) }
-c'i/ 1-m2 bVi -m2 cos (0 -00)-(b" +Ma') sin ((3- 0) t .
The coefficient of the cosine is (b'- ma")(c + m) - bc' (1 - M2), which is
b'c- bc'+ m (b'- ca") + m2 (- a"+ bc'), a" + m (b'- ca") + m2 (- b'c), -

(1 + cm)(-a" + mb'),
 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature. 97

and similarly the coefficient of V 1- m2 sine is - c" (b'- ma") + c' (b" + ma'),
- b'c"+ b"c'+ m (a'c'+ a"c"), - a + m (- ac), =(1 + cm)(- a). Cal-
culating in like manner the values of y and z, and putting for shortness
X1= (-a" + rnb') cos (Q3-80)- aV 1-m2 sin ( 3-0),
Y1-= b"-ma') cos (Q3-0) -bV I -- ml sin (-? 60),
Z=- c"V1 -in2) cos (E0- 00) + (c + i) sinQ(3- o-
we have
x= qV M2IX+ XiV 0 + (m2l 1) q)2- m (b'- ma")/0 + (m 1) q2,
y __= pVi-m2Y+ YiV0 + (m2- 1p2+m (a'+ mb")VO+ (VI2_>2,
1)
Z-VI - m2{pV1-,m2Z+ Z1VO+( n2 1)q

which are the requiired expressions of x, y, z in termnsof t and 0. It will be

noticed that X, X1, Y, Y1, Z, Z1, each contain a term with cos (9 - e0) and
one with sin (0 - 0,), but as the terms in X1, YI1,Z1 are each multiplied by
V/0 + (m2- 1) c2, the cosinieand sine terms of X, X1, of Y, Y1and of Z, Z1 do

not in any case unite into a single term.

I remark that we have identically
aX+ bY + cVI qn2Z = 0,
-

aXafl+ b
bY1+
Y1+cCV/1- _M2Zl
IZ= 0.
The foregoing values of x, y, z thus satisfy ax + by + cz = 0, which is one
of the six equations. The others of them might be verified without diffi-
culty. I recall that we have a, b, c- (a' + mb"), -(b'- ma"), -c'; the
six equations might therefore be written
A2+B2+ C2=1,
x +- (b' -ma") y + c'z = 0,
(a' + mbb")
(a' + mb")A + (b' - ma")B + c' C = - c'm
x2+ y2 + (Z_ M)2 =0 + m2p2,
Ax+By+C(z -imp) . 1
Adx+Bdy+ Cdz 0.

THE CASE PS10 = SERRET'SFIRST CASE OF PS.

This is at once deduced from PS30 by writing therein Mn= 0; the formulae
are a good deal more simple. We introduce as before the rectangularcoefficients
13
98 CAYLEY: On the Surfaces with Plane or Spherical Curvesof Curvature.

a, b, c, a', b', c', a", b", c", and the values of a, b, c then are a', b', c'. The six
equations, using therein these values for a, b, c, are
A2 +B2 + C0 = 1,
a'x + by + c'z = 0,
a'A+b'B +c'C=0-,
X2+ y2 + z2 = 0
Ax + By + Cz =.
Adx + Bdy + Cdz= O.

The function CDis such that p = , -F, F2- 4Y We have

,I- + 4041)2
4CFCF' V/M

e=
sin (3 - ,/ cos -P-

sin8(0=_ cos80= vo/)0

and thence
sin (O -() 24VVOjf (DI cos(?- - I -
0)=
Also
sinc =
C 42
V-c2
/ - C2
; sin T=
c ci
4 2'
=-c2 i c2,Cos T=
; - T- O0+ C=
C c cos (O- 0) -c/ sin (8 -890),
C
VI - _ C2 c" cos (8 - 80) + c sin (O--0).
We have
A =X = a cos (8 - 80)- a" sin (? - 80); X1 = a" cos (- -0) + a sin (e-0),
B= Y= b cos(8 -?0)- bsin(0- -0); Y =b"cosQ(3- -0)+bsin(8- 8o),
0-= Y= c cos(E- 0)- c sin ((3 '0); Z1 =c" cos(0 -90)+c sin (0-?- ),
and then
X-Xq + X1V0-(P2,
Y= p + Y1V0- V2,
=
Z=Zq +ZVl0-p 0 ,

which are the expressions of the coordinates in terms of the parameters t and 0.
(To be continuqd.)