LECTURE 10

3.3 First fundamental form, metric and area

For curves, we used the structure given by ambient space of R3 to define length, curvature,
torsion and the Frenet frame (when the ambient space is R2 we have a bit more structure
and we can give define a signed notion of curvature). We now study similarly notions of
area, curvature and tangent space inherited from the ambient space R3 for surfaces. The
Riemannian metric is the local notion of what measures distance, it is the surface equivalent
to the length for our curve, and in fact can be defined for higher-dimensional objects. It was
however first called first fundamental form when invented for surface, and this name
survived in this particular case. Later, we will look at second order objects (like n and n in
the case of curves) which will allow us to define notions of curvature.

Definition 84. A surface S is a 2-manifold in RN , N ≥ 3.

Remark 85. In this course we will restrict to the case when the ambient space has dimension
N = 3. However not all surfaces can be embedded in R3 , check e.g. Klein’s bottle online!

Definition 86. Given S ⊂ R3 a surface, the first fundamental form Ip at p ∈ S is

the quadratic form on the tangent space Tp S ⊂ R3 (which is a 2-dimensional vector space)
induced by the inner product of R3 . That means that if v ∈ Tp S then Ip (v, v) = v · v = |v|2 .

Remark 87. So this definition means that that locally the metric used on the tangent space
is simply inherited from the inner product in the ambient space R3 . This local metric will
allow us to do various measurements on the surface, like length and area. But the mistake to
avoid, as in all this chapter, is that this “trivial inner product” on Tp S does not correspond
to a trivial inner product in R2 when parametrising the surface by open sets of R2 . The
parametrisation will capture the deformations of the surface and encode them in all the
trivial inner product of R3 is deformed when “brought back” to an inner product in R2 .
Remark 88. (Non-examinable) For a general manifold X (defined along the intrinsic viewpoint
at the end of the previous chapter), a Riemannian metric on X is a smooth map g mapping
each p ∈ X to a positive quadratic form on Tp X (allowing non-positive quadratic forms would
lead to Lorentzian geometry). A natural way to construct new Riemannian metrics is by
“pull-back”: if f : X → Y immersion between manifolds and Y has a Riemannian metric g,
then hp (w) := gf (p) (Dfp (w)) defines a Riemannian metric on X. In this viewpoint, the first
fundamental form is the Riemannian metric on S = X determined by the inner product in
R3 = Y , where f is simply the canonical inclusion map S → R3 .
Since we have a local notion of metric, we can define a corresponding notion of isometry:

Definition 89. Given two surfaces S1 and S2 in R3 , a smooth map f : S1 → S2 is isometric

if for all p ∈ S1 , the differential Dfp : Tp S1 → Tf (p) S2 is a linear isometry.

In accordance with the remark above about the mistake to avoid, let us now express the
first fundamental form (Riemannian metric) locally, i.e. in a given parametrisation.

22
Proposition 90. Given S ⊂ R3 a surface, a point p ∈ S on the surface, and a parametrisa-
tion ϕ : U → U around p, i.e. ϕ = ϕ(u, v) mapping an open set in R2 to an open set around
p in S, with say ϕ(u0 , v0 ) = p. Then the first fundamental form at p is given, for w ∈ Tp S

2


2
Ip (w) = |w|2 = E P∂u ϕ(u0 ,v0 ) w + 2F P∂u ϕ(u0 ,v0 ) w P∂v ϕ(u0 ,v0 ) w + G P∂v ϕ(u0 ,v0 ) w

where P∂u ϕ(u0 ,v0 ) w and P∂u ϕ(u0 ,v0 ) w denote the unique decomposition of w in the basis of
vectors ∂u ϕ(u0 , v0 ) and ∂v ϕ(u0 , v0 ) of Tp S, and with the following formulas for the coefficients


 E := ∂u ϕ(u0 , v0 ) · ∂u ϕ(u0 , v0 ),


F := ∂u ϕ(u0 , v0 ) · ∂v ϕ(u0 , v0 ),




G := ∂v ϕ(u0 , v0 ) · ∂v ϕ(u0 , v0 ).

Proof. Given w ∈ Tp S we can always consider a curve α : I → S so that α(0) = p and α̇(0) =
w. Then, assuming without loss of generality that this curve is included in the neighbourhood
U of our parametrisation, we find t 7→ (u(t), v(t)) smooth so that ϕ(u(t), v(t)) = α(t) (by
inverting ϕ). The chain rule then gives

w = α̇(0) = [∂u ϕ(u0 , v0 )] u̇(0) + [∂v ϕ(u0 , v0 )] v̇(0)

which means that (u̇(0), v̇(0)) is the decomposition of w in the basis (∂u ϕ(u0 , v0 ), ∂v ϕ(u0 , v0 )),
and this gives the result by taking the norm (note that (∂u ϕ(u0 , v0 ), ∂v ϕ(u0 , v0 )) is linearly
independent because ϕ diffeomorphism, and that it spans Tp S by definition of Tp S).

Remark 91. Note that the basis (∂u ϕ(u0 , v0 ), ∂v ϕ(u0 , v0 )) of the tangent space created by
the parametrisation is not necessarily orthogonal or normalised, which explains the need for
coefficients and a cross-term. We will come back later to how to build nicer parametrisations.
Example 92. Consider the 2-sphere S centred at zero and with radius r > 0 in R3 . A
parametrisation around Cambridge C = ϕ(π/4, 0) is, for u ∈ (−π/2, π/2) and v ∈ (−π, π)

ϕ(u, v) = (r cos v cos u, r sin v cos u, r sin u).

This gives the following basis for the tangent space at C

   
r r r
∂u ϕ|(π/4,0) = − √ , 0, √ , ∂v ϕ|(π/4,0) = 0, √ , 0 .
2 2 2
The first fundamental form is then (exercise), for w ∈ TC S ⊂ R3 :
(w1 − w3 )2 + w22
IC (w) = r2 .
2
Example 93. Consider S the torus obtained by rotating a circle of radius r > 0 about a
straight line belonging to the plane of the circle and at a distance a > r away from the centre
of the circle. A parametrization which covers all of the torus except for a meridian and a
parallel is given by

ϕ(u, v) = ((a + r cos u) cos v, (a + r cos u) sin v, r sin u)

where u, v ∈ (0, 2π). Then we have E = r2 , F = 0 and G = (r cos u + a)2 .

23
 We can now compute length and area thanks to the first fundamental form. The next
proposition on length confirms that the first fundamental form is a local notion of metric.
Proposition 94. Given S ⊂ R3 a surface and α : I → S a curve on the surface included in
a parametrisation ϕ, then the length of the curve
Z q
ℓ(α) = Iα(t) (α̇(t)) dt.
I

Proof. This is in fact trivial since Iα(t) (α̇(t)) = |α̇(t)|2 .

We now turn to the area. The local basis on the tangent space created by a parametrisa-
tion allows to measure an infinitesimal element of area.
Definition 95. Given S ⊂ R3 a surface and Ω ⊂ S a bounded connected non-empty open
subset (bounded means that Ω is included in a ball of R3 ) that is included in the image of a
parametrisation ϕ : U → U with U open in R2 and Ω ⊂ U ⊂ S, then the positive number
Z
A(Ω) := |∂u ϕ × ∂v ϕ| du dv
ϕ−1 (Ω)

is called the area of Ω (note that it is an integral on an open subset of R2 ).

This definition requires some proof to make sure the quantity we have defined is not
parametrisation-dependent.
Proposition 96. The area defined above is independent of the parametrisation ϕ√(provided
its image contains Ω). Moreover given such a parametrisation, |∂u ϕ × ∂v ϕ| = EG − F 2
where E, F and G are the coefficients of the first fundamental form, hence the area can be
expressed in terms of the first fundamental form written in this parametrisation:
Z p Z q
A(Ω) := 2
EG − F du dv = det I(u,v) du dv.
ϕ−1 (Ω) ϕ−1 (Ω)

Proof. Given two parametrisations ϕ : U → U and ψ : V → V with Ω ⊂ U ∩ V, we want to

prove that the area defined as above is the same
Z Z
|∂u ϕ × ∂v ϕ| du dv = |∂ũ ψ × ∂ṽ ψ| dũ dṽ.
ϕ−1 (Ω) ψ −1 (Ω)

Let us denote the transition map by θ := ϕ−1 ◦ ψ between open sets of R2 , then we have

|∂ũ ψ × ∂ṽ ψ| = |Jθ (ũ, ṽ)| |∂u ϕ × ∂v ϕ| ◦ θ(ũ, ṽ)

where Jθ (ũ, ṽ) = det Dθ(ũ,ṽ) is the Jacobian determinant of θ. Indeed by the chain rule

∂ũ ψ = [(∂u ϕ) ◦ θ] (∂ũ θ)1 + [(∂v ϕ) ◦ θ] (∂ũ θ)2

∂ṽ ψ = [(∂u ϕ) ◦ θ] (∂ṽ θ)1 + [(∂v ϕ) ◦ θ] (∂ṽ θ)2

so that ∂ũ ψ × ∂ṽ ψ = (∂u ϕ × ∂v ϕ) ◦ θ((∂ũ θ)1 (∂ṽ θ)2 − (∂ũ θ)2 (∂ṽ θ)1 ) and the claimed formula.
The proposition follows by the change of variable (u, v) = θ(ũ, ṽ). The formulas in terms of
the first fundamental form are immediate.

24