Skip to content
Topos Uranos

Summary

This lesson measures solids with the method of the unit. Since the volume of a solid bounded by curved surfaces is not yet defined, it lays down three requirements (a cylinder, right or oblique, measures the area of its base times its height; a solid contained in another has no greater volume; a solid decomposed into two that do not overlap measures the sum of the two) and proves that, if the cross-sections of a solid by the planes perpendicular to an axis have an integrable area and each slice lies between two cylinders whose bases differ little, the only volume compatible with them is the integral of the cross-sectional area. That number is adopted as the definition, and Cavalieri’s principle is a corollary. With oblique cylinders one obtains the volume of every cone whose base is a convex figure lying between two continuous graphs, in particular that of the pyramid with a convex base, whose formula analytic geometry in space admitted. Solids of revolution are measured with disks and washers, and also with cylindrical shells; for a continuous, nonnegative and strictly monotonic function, the two descriptions of the same solid give the same number, and the proof uses no derivatives: only a summation by parts and the bounds for the integral.

Learning objectives

  1. Explain why volume must be defined, state its requirements and prove that the volume of a solid with cross-sections of integrable area that satisfies the cylinder condition is the integral of that area.
  2. Compute volumes with Cavalieri’s principle, prove the formula for the volume of the pyramid that analytic geometry admitted and deduce those of the cone and of the ball.
  3. Compute volumes of solids of revolution by disks, washers and cylindrical shells, and prove that the two methods agree for strictly monotonic functions.
  4. Distinguish in each formula what is definition and what is theorem, and recognize which hypotheses cannot be omitted.

Volume and its requirements

The lesson on the area problem did not define the area of a curved figure: it laid down three requirements and showed that they suffice to determine some areas. The lesson on areas and the average value of a function made the method of the unit explicit (requirements, simple pieces whose measure is known, lower and upper sums, passage to the limit with the integrability criterion or with Darboux’s theorem), and this lesson applies it to volume, which is not defined either for a solid bounded by curved surfaces. The simple pieces are now cylinders, and their bases are the plane figures that already have an area: polygons, with the areas admitted in the lesson on the area problem; regions between two graphs of integrable functions, with respect to either of the two axes of the plane, with the area defined in the lesson on areas and the average value; and unions of finitely many of those figures that do not overlap, with the sum of their areas. A point or a line segment has area 00, like the rectangles with a zero side in the lesson on the area problem. When a figure admits several of those descriptions, it is admitted that all of them give it the same area, a fact which, as the lesson on areas and the average value said, is proved in later courses; for the polygons and disks this lesson uses, the consistency is checked where they appear.

Requirements for volume. A volume assignment associates with certain solids, sets of points in space, a number greater than or equal to 00, their volume, with the three properties below.
  1. Cylinder: let BB be a figure with area contained in a plane x=ux = u, a number w>uw > u and a vector (1,σ,τ)(1, \sigma, \tau). The cylinder with base BB, height w−uw - u and generators parallel to that vector, the set of the points (x,y+σ (x−u),z+τ (x−u))\left(x, y + \sigma \, (x - u), z + \tau \, (x - u)\right) with (u,y,z)(u, y, z) in BB and u≤x≤wu \leq x \leq w, has volume equal to the area of BB times w−uw - u. It is right if σ=τ=0\sigma = \tau = 0, and oblique otherwise. The same holds with the roles of the three coordinates permuted.
  2. Monotonicity: if a solid is contained in another and both have a volume, the volume of the first is not greater than that of the second.
  3. Additivity: if a solid is the union of two solids with volume that do not overlap, that is, that have in common only points of their boundaries, then it has a volume, and its volume is the sum of the volumes of the two.

By induction on the number of pieces, additivity holds for the union of any finite number of solids with volume that pairwise do not overlap. The volumes of elementary geometry satisfy the three requirements; for example, a prism, right or oblique, with its bases in two planes perpendicular to an axis, is a cylinder whose base is a polygon. The requirements do not include that a translated or rotated solid keeps its volume, and this lesson does not use it: each solid is placed with respect to the coordinate axes. For a solid EE and a number xx, the cross-section ExE_x is the set of the points (y,z)(y, z) such that (x,y,z)(x, y, z) lies in EE: the figure that the plane perpendicular to the axis of abscissas through xx cuts out of the solid.

TheoremVolume is the integral of the cross-sectional area

Let EE be a solid contained between the planes x=ax = a and x=bx = b, with a<ba < b, whose cross-section ExE_x has, for each xx in [a,b]\icc{a}{b}, an area A(x)A(x), with AA integrable on [a,b]\icc{a}{b}. Suppose that EE satisfies the cylinder condition: for each ε>0\varepsilon > 0 there are a partition P={x0,…,xn}P = \{x_0, \ldots, x_n\} of [a,b]\icc{a}{b} and numbers αk≥0\alpha_k \geq 0 and βk\beta_k, for k=1,…,nk = 1, \ldots, n, such that the portion of EE between the planes x=xk−1x = x_{k - 1} and x=xkx = x_k contains a cylinder of height Δxk\Delta x_k whose base has area αk\alpha_k (except if αk=0\alpha_k = 0) and is contained in a cylinder of height Δxk\Delta x_k whose base has area βk\beta_k (both with the base in the plane x=xk−1x = x_{k - 1}), with αk≤A(x)≤βk\alpha_k \leq A(x) \leq \beta_k for every xx in [xk−1,xk]\icc{x_{k - 1}}{x_k} and with ∑k=1n(βk−αk) Δxk<ε\sum_{k=1}^{n} (\beta_k - \alpha_k) \, \Delta x_k < \varepsilon. If EE has a volume VV that satisfies the requirements for volume, then:

V=∫abA(x) dxV = \int_a^b A(x) \, \mathrm{d}x

The cylinders contained in the slices lie inside the solid, and those that contain them cover it; their volumes enclose both the volume and the integral, in an interval as short as desired.

Proof

  1. ∑k=1nαk Δxk≤V≤∑k=1nβk Δxk\dato{vol}{\sum_{k=1}^{n} \alpha_k \, \Delta x_k \leq V \leq \sum_{k=1}^{n} \beta_k \, \Delta x_k}

    Let ε>0\varepsilon > 0 and PP, αk\alpha_k and βk\beta_k be as in the condition. The cylinders contained in the portions of EE with αk>0\alpha_k > 0 have in common only points of the planes x=xkx = x_k, which lie on their boundaries; by the cylinder and additivity requirements, the latter extended to several pieces, their union has volume ∑αk Δxk\sum \alpha_k \, \Delta x_k, and by monotonicity it does not exceed VV (the terms with αk=0\alpha_k = 0 contribute nothing, and if all of them are such, the bound holds because V≥0V \geq 0). Each point of EE lies in one of the portions, so EE is contained in the union of the cylinders that contain them, which do not overlap either, because each lies between the two planes of its portion, and the same requirements give the upper bound.

  2. ∑k=1nαk Δxk≤L(A,P)≤∫abA≤U(A,P)≤∑k=1nβk Δxk\dato{int}{\sum_{k=1}^{n} \alpha_k \, \Delta x_k \leq L(A, P) \leq \int_a^b A \leq U(A, P) \leq \sum_{k=1}^{n} \beta_k \, \Delta x_k}

    The function AA is integrable, and therefore bounded; by the completeness axiom, from the lesson on number systems, it has on each subinterval an infimum mkm_k and a supremum MkM_k. Since αk\alpha_k is a lower bound of the values of AA on the subinterval and βk\beta_k an upper bound, αk≤mk\alpha_k \leq m_k and Mk≤βkM_k \leq \beta_k. By the proposition on the order of the lower and upper integrals, from the lesson on the Riemann integral:

  3. (∀ε>0)(∣V−∫abA∣<ε)→V=∫abA(x) dx(\forall \varepsilon > 0)\left(\abs{V - \int_a^b A} < \varepsilon\right) \rightarrow \resaltar{V = \int_a^b A(x) \, \mathrm{d}x}

    The volume and the integral lie in one and the same interval, of length less than ε\varepsilon, and two numbers of an interval are at most its length apart. Since ε\varepsilon is arbitrary, if the distance between them were positive, taking it as ε\varepsilon it would turn out less than itself; therefore it is 00.

DefinitionVolume of a solid

Let EE be a solid contained between the planes x=ax = a and x=bx = b, with a<ba < b, whose cross-sections have areas A(x)A(x), with AA integrable on [a,b]\icc{a}{b}, and which satisfies the cylinder condition. The volume of EE is ∫abA(x) dx\int_a^b A(x) \, \mathrm{d}x. The same holds with the cross-sections perpendicular to the axis of ordinates or to that of the third coordinate.

By the theorem, it is the only value compatible with the requirements, as in the definition of the area under a graph, from the lesson on the Riemann integral, and it is adopted as the definition for the same reason: the theorem only says which number the volume can be, and the definition ensures that there is one. The cylinder requirement is satisfied: the cross-section of a cylinder by each plane between its bases is the translated base, which has the same area: for a polygon, by elementary geometry, and for a region between two graphs, by the lemma on the translation and reflection of the integral, from the lesson on techniques for definite integrals; and each slice of the cylinder is a cylinder whose base is one of those cross-sections, so the condition holds with αk=βk\alpha_k = \beta_k. The lesson does not prove, on the other hand, that the definition satisfies the requirements for all solids at once: one and the same solid can be cut by planes perpendicular to two different axes, and the definition would in principle give it two numbers. The theorem on the equivalence of washers and shells, further below, and the proposed problem “The wedge of Archimedes” show that they agree in important cases; the general case is a topic for later courses, where integrals in several variables are studied, and nothing in this lesson depends on it.

CorollaryNested cross-sections

Let EE be a solid contained between the planes x=ax = a and x=bx = b, with a<ba < b, whose cross-sections have areas A(x)A(x), with AA integrable on [a,b]\icc{a}{b}. Suppose that in each interval [u,w]\icc{u}{w} contained in [a,b]\icc{a}{b}, with u<wu < w, there are two points x′x' and x′′x'' of that interval such that, for every xx in [u,w]\icc{u}{w}, Ex′⊆Ex⊆Ex′′E_{x'} \subseteq E_x \subseteq E_{x''} and A(x′)≤A(x)≤A(x′′)A(x') \leq A(x) \leq A(x''). Then EE satisfies the cylinder condition. This happens, for example, if the cross-sections grow with xx and so do their areas: it suffices to take x′=ux' = u and x′′=wx'' = w. When the cross-sections are nested, the right cylinders over the smallest and the largest give exactly the lower and upper sums of AA.

Proof

  1. ∑k=1n(βk−αk) Δxk=∑k=1n(Mk−mk) Δxk=U(A,P)−L(A,P)<ε\sum_{k=1}^{n} (\beta_k - \alpha_k) \, \Delta x_k = \sum_{k=1}^{n} (M_k - m_k) \, \Delta x_k = U(A, P) - L(A, P) < \resaltar{\varepsilon}

    Let ε>0\varepsilon > 0 and, by Riemann’s integrability criterion, from the lesson on the Riemann integral, a partition PP with U(A,P)−L(A,P)<εU(A, P) - L(A, P) < \varepsilon. On each subinterval, the right cylinder with base Ex′E_{x'} between the planes x=xk−1x = x_{k - 1} and x=xkx = x_k is contained in EE, because each of its cross-sections, Ex′E_{x'}, is contained in that of EE; likewise, the portion of EE is contained in the right cylinder with base Ex′′E_{x''}. Let αk=A(x′)\alpha_k = A(x') and βk=A(x′′)\beta_k = A(x''): they are the minimum and the maximum of AA on the subinterval, which therefore coincide with mkm_k and MkM_k.

CorollaryCavalieri’s principle

Let E1E_1 and E2E_2 be solids contained between the planes x=ax = a and x=bx = b, with a<ba < b, whose cross-sections have areas A1(x)A_1(x) and A2(x)A_2(x) integrable on [a,b]\icc{a}{b}, and which satisfy the cylinder condition. If there is a constant λ>0\lambda > 0 such that A1(x)=λ A2(x)A_1(x) = \lambda \, A_2(x) for every xx in [a,b]\icc{a}{b}, the volume of E1E_1 is λ\lambda times that of E2E_2; in particular, if the cross-sections of the two solids by each plane have the same area, the two have the same volume. Two stacks of the same coins, one upright and the other slanted, have the same volume.

Proof

  1. V1=∫abA1=∫abλ A2=λ∫abA2=λ V2V_1 = \int_a^b A_1 = \int_a^b \lambda \, A_2 = \lambda \int_a^b A_2 = \resaltar{\lambda \, V_2}

    By the definition of the volume of a solid, the two volumes are the integrals of the cross-sectional areas, and by the linearity of the integral, from the lesson on properties of the definite integral, the constant comes out of the integral.

Cavalieri stated it in 1635 as a principle of indivisibles, without proof; here it is a corollary of the definition, and its content lies in the hypothesis: the two solids must satisfy the cylinder condition. Thanks to the oblique cylinders of the requirements, inclined solids satisfy it too, such as the cones treated below.