1.5.

The Second Fundamental Form 1

1.5 The Second Fundamental Form

Notation. We adopt the Einstein summation convention in which any expression

that has a single index appearing both as a subscript and a superscript is assumed
to be summed over that index.

~ i as viX
~ i.
X
Example. We denote vi X
i

X
Example. We denote gij vi w j as gij vi w j .
i,j

Note. We have treated a path α

~ (t) along a surface M as if it were the trajectory of
~ 00 (t) as the acceleration of the particle. Well,
a particle in E 3 . We then interprete α
a particle can accelerate in two different ways: (1) it can accelerate in the direction
of travel, and (2) it can accelerate by changing its direction of travel. We can
~ 00 into two components, α
therefore decompose α ~ 00T~ (representing acceleration in the
~ 00N~ (representing acceleration that changes the direction of
direction of travel) and α
~ 00T~ as the component
travel). You may have dealt with this in Calculus 3 by taking α
~ 00 in the direction of α
of α ~ 0 , computed as
0
  0
α
~ α
~
~ 00T~ = α
α ~ 00 · 0
k~
α k k~α0 k
~ 00N~ as the “remaining component” of α
and α ~ (that is, α ~ 00N~ = α
~ 00 − α
~ 00T~ ). This is
reminiscent of the Frenet formulas or the Frenet frame (T~ , N
~ , B)
~ of Exercise 1.1.14).
1.5. The Second Fundamental Form 2

Note. With α~ parameterized in terms of arc length s, α~ =α ~ 1 (s), u2(s))

~ (s) = X(u
we have the unit tangent vector T~ (s) = α ~ i. We saw in Section 1.1
~ 0 (s) = ui0 X
~ 00 (s) = T~ 0 (s) is a vector normal to α
that α ~ 0 (T~ 0 = kN
~ - see Exercise 1.1.14).
~ 00 into two orthogonal components, but this
In this section, we again decompose α
time we make explicit use of the surface M . We wish to write

~ 00 = α
α ~ 00tan + α
~ 00nor

~ 00tan is the component of α

where α ~ 00 tangent to M and α
~ 00nor is the component of α ~ 00
normal to M . Notice that α ~ 00tan will be a linear combination of X ~ 1 and X
~ 2 (they
are a basis for the tangent plane, recall) and α~ 00nor will be a multiple of the unit
~ ~
normal vector to M , U ~ = X1 × X2 ).
~ (calculated as U
kX~1 × X ~ 2k

~ 00 , α
In the figure, α ~ 00tan , and α
~ 00nor all line in the same plane.
1.5. The Second Fundamental Form 3

Note. Since α ~ 1 (s), u2 (s)) and α

~ (s) = X(u ~ i (here, 0 means d/ds), then
~ 0 = ui0 X

dX~i
~ i + ui0 X
~ 00 = ui00 X
α ~ 0 = ui00 X
~ i + ui0 .
i
ds
~ i is part of α
Now ui00 X ~ 0 may also have a component in the tangent
~ 00tan , but ui0 X i
plane. Well,

dX~i d h~ 1 i ∂X~ i du1 ∂ X

~ i du2
2
= Xi(u (s), u (s)) = + 2
ds ds ∂u1 ds ∂u ds
∂X~i ~i
∂X
10
= 1
u + 2
u20
∂u ∂u
(notice that we sum over j in the last term, since we treat the partial derivative
∂ 2X~
j
with respect to u as if j were a subscript). If we denote ~ ij (we have
=X
i
∂u ∂u j
assumed continuous second partials, so the order of differentiation doesn’t matter)
dX~i
then we have ~ ij uj0 . So acceleration becomes
=X
ds
~ r + ui0 uj0 X
~ 00 = ur00 X
α ~ ij .

We now need only to write X ~ ij in terms of a component in the tangent plane (and
so in terms of X~ 1 and X~ 2 ) and a component normal to the tangent plane (which
~ ).
will be a multiple of U

Definition. With the notation above, we define the formulae of Gauss as

~ ij = Γr X
X ~ ~
ij r + Lij U .

That is we define Lij as the projection of X~ ij in the direction U~ . Notice, how-

ever, that Γrij may not be the projection of X~ ij onto X
~ r since the X~ r ’s are not
orthonormal.
1.5. The Second Fundamental Form 4

Note. Since projections are computed from dot products, we immediately have
that
~1 × X
X ~2
~ ~ ~
Lij = Xij · U = Xij · .
~1 × X
kX ~ 2k

Note. We therefore have

~ 00 = α
~ 00tan + α
~ 00nor = ur00 + Γrij ui0 uj0 X~ r + Lij ui0 uj0 U~.



α

Definition. The second fundamental form of surface M is the matrix

!
L11 L12
.
L21 L22

(Notice this differs from the text’s definition on page 44.) We denote the determi-
nate of this matrix as L.

Note. The second fundamental form is a function of u = u1 and v = u2 . Also,

~ 12 = X
since we have X ~ 21, it follows that L12 = L21 and so (Lij ) is a symmetric
matrix.

Note. We will see that the second fundamental form reflects the extrinsic geometry
of surface M (that is, the way M is embedded in E 3; “how it curves relative to
that space” as the text says).
1.5. The Second Fundamental Form 5

Example (Exercise 1.5.2). Compute L from the second fundamental form of

the surface of revolution

~
X(u, v) = (f (u) cos v, f (u) sin v, g(u)).

Solution. Well
~
~ 1 = ∂ X = (f 0(u) cos v, f 0 (u) sin v, g 0 (u))
X
∂u
~
~ 2 = ∂ X = (−f (u) sin v, f (u) cos v, 0)
X
∂v
and so (from Exercise 1.4.5)

~ = f (u)
U p (−g 0 (u) cos v, −g 0 (u) sin v, f 0 (u)).
0 2 0
|f (u)| f (u) + g (u) 2

Next,
∂ 2X~
~
X11 = = (f 00 (u) cos v, f 00 (u) sin v, g 00 (u))
∂ u
2
2~
~ 22 = ∂ X = (−f (u) cos v, −f (u) sin v, 0)
X
∂ 2v
∂ 2X~
~
X12 = = (−f 0 (u) sin v, f 0 (u) cos v, 0) = X ~ 21.
∂u∂v
So

~ 11 · U
~ = f (u)
L11 = X p (−f 00 (u)g 0(u) cos2 v
|f (u)| f 0(u)2 + g 0 (u)2
−f 00 (u)g 0 (u) sin2 v + f 0(u)g 00 (u))
f (u)(f 0 (u)g 00 (u) − f 00 (u)g 0(u))
= p
|f (u)| f 0 (u)2 + g 0 (u)2
~ ~ (f (u)g 0 (u) cos v sin v − f (u)g 0 (u) cos v sin v + 0)f (u)
L12 = X12 · U = p
|f (u)| f 0(u)2 + g 0 (u)2
= 0 = L21
1.5. The Second Fundamental Form 6

~ 22 · U
~ = f (u)
L22 = X p (f (u)g 0 (u) cos2 v
|f (u)| f 0(u)2 + g 0 (u)
+f (u)g 0 (u) sin2 v + 0)
f (u)2 g 0(u) |f (u)|g 0(u)
= p =p .
|f (u)| f 0 (u) + g 0 (u)2 f 0 (u)2 + g 0 (u)2
Therefore we have

L = det Lij = L11 L22 − L12 L21

f (u)(f 0(u)g 00 (u) − f 00 (u)g 0 (u)) |f (u)|g 0(u)
= p p
|f (u)| f 0 (u)2 + g 0(u)2 f 0 (u)2 + g 0 (u)2
f (u)g 0(u)(f 0 (u)g 00 (u) − f 00 (u)g 0(u))
= .
f 0(u)2 + g 0 (u)2

Definition. Let ~v = viX ~ i be a unit vector tangent to M at P~ . The normal

curvature of M at P~ in the direction ~v, denoted kn (~v) is

kn (~v) = Lij vi vj

where ~v = (v1 , v2 ) (the coordinate vector of ~v with respect to the ordered bases
~ 1, X
(X ~ 2 ) of the tangent plane).

Example (Exercise 1.5.5). Find the normal curvature of the surface z = f (x, y)
at an arbitrary point, in the direction of a unit tangent vector (a, b, c) at that point.

Solution. We have
~
~ 1 = ∂ X = (1, 0, ∂f (u, v)) = (1, 0, fu )
X
∂u ∂u
~
~ 2 = ∂ X = (0, 1, ∂f (u, v)) = (0, 1, fv ).
X
∂v ∂v
1.5. The Second Fundamental Form 7

So
~i ~j ~k
~1 × X
X ~2 = 1 0 fu = (−fu , −fv , 1)
0 1 fv
~1 × X
~ 2k =
p
and kX (fu)2 + (fv )2 + 1. Therefore
~ ~
~ = X1 × X2 = p
U
1
(−fu , −fv , 1).
~1 × X
kX ~ 2k (fu)2 + (fv )2 + 1
Now
∂ 2X~
~
X11 = = (0, 0, fuu )
∂u2
∂ 2X~
~
X12 = ~ 21
= (0, 0, fuv ) = X
∂u ∂v
∂ 2X~
~
X22 = = (0, 0, fvv )
∂v2
and so

~ 11 · U
~ =p fuu
L11 = X
(fu)2 + (fv )2 + 1
~ 22 · U
~ =p fvv
L22 = X
(fu)2 + (fv )2 + 1
~ 12 · U
~ =p fuv
L12 = X = L21 .
(fu)2 + (fv )2 + 1
~ i = v1 (1, 0, fu ) + v2 (0, 1, fv ) = (a, b, c), implying that v1 = a and
Now ~v = viX
v2 = b. Hence

kn (~v ) = Lij vivj

= L11v1 v1 + 2L12 v1 v2 + L22 v2 v2
1
= p (a2fuu + 2abfuv + b2 fvv ).
(fu)2 + (fv )2 + 1
1.5. The Second Fundamental Form 8

Note. If α~ = X(u~ 1 (s), u2(s)) is a curve on M , P~ is a point on M with α ~ (s0) = P~

~ 0 (s0) then α
and ~v = α ~ i(u1 (s0), u2 (s0)) and so vi = ui0 (s0) (see page
~ 0 (s0) = ui0 (s0)X
35 for representation of a tangent vector: ~v = vi X ~ i). Therefore

kn (~v) = Lij vi vj = Lij ui0 uj0 .

~1 × X
X ~2
Now α 00 ~ r00i0 j0 ~ ~
~ = u Xr + u u Xij (equation (16), page 43), and U = so
~1 × X
kX ~ 2k
  X ~1 × X ~2
00
α ~ =
~ ·U ~r00 i0 j0 ~
u Xr + u u Xij ·
kX~1 × X ~ k
! 2
~ ~
= ~0 + ui0 uj0 X~ ij · X1 × X2 = ui0 uj0 Lij .
~ ~
kX 1 × X 2 k

~ . This equation is used in Exercise 1.5.6.

~ 00 · U
Hence kn (~v ) = α

Revised: 6/13/2019