Professional Documents
Culture Documents
Textbook Introduction To Complex Theory of Differential Equations Savin Ebook All Chapter PDF
Textbook Introduction To Complex Theory of Differential Equations Savin Ebook All Chapter PDF
https://textbookfull.com/product/introduction-to-partial-
differential-equations-peter-j-olver/
https://textbookfull.com/product/distribution-theory-applied-to-
differential-equations-1st-edition-adina-chirila/
https://textbookfull.com/product/a-modern-introduction-to-
differential-equations-3rd-edition-henry-j-ricardo/
https://textbookfull.com/product/introduction-to-the-theory-of-
complex-systems-stefan-thurner/
Ordinary differential equations : an introduction to
the fundamentals Second Edition. Edition Howell
https://textbookfull.com/product/ordinary-differential-equations-
an-introduction-to-the-fundamentals-second-edition-edition-
howell/
https://textbookfull.com/product/numerical-solution-of-
differential-equations-introduction-to-finite-difference-and-
finite-element-methods-1st-edition-zhilin-li/
https://textbookfull.com/product/theory-of-differential-
equations-in-engineering-and-mechanics-1st-edition-kam-tim-chau/
https://textbookfull.com/product/modelling-with-ordinary-
differential-equations-dreyer/
https://textbookfull.com/product/elementary-differential-
equations-second-edition-roberts/
Anton Savin • Boris Sternin
and and
Mathematics Subject Classification (2010): 58-02, 58J32, 58Z05, 58J47, 35Gxx, 32W50, 86-XX, 78-XX
v
vi Preface
of the Cauchy problem and describe the singularities of the solution. Note that solutions
of complex Cauchy problems always have singularities (possibly located at infinity), and
therefore the function classes, in which the Cauchy problem is considered, are the classes
of ramified analytic functions. Chapter 8 is devoted to studying Cauchy problems for
equations with variable coefficients using Leray’s uniformization method. The results ob-
tained here are valid in the small, even though presently there is an apparatus that permits
us to construct asymptotics in the large, i.e., far from the original manifold. Unfortunately,
this apparatus is far more complicated technically and its exposition is beyond the scope
of this book. The final Chapters 9 and 10 are devoted to applications: to the solution of
Poincaré’s balayage problem mentioned above as well as to an effective construction of
“mother bodies.”
At the end of each chapter we give bibliographic remarks on the references related
to that chapter. We should mention that during the preparation of this book we widely
used classical works by J. Leray and also the works by B.Yu. Sternin and V.E. Shatalov
and their coauthors on complex theory of differential equations. These are briefly the
contents of the book.
Acknowledgments. The results discussed in this book were delivered at a number of
scientific seminars [seminars of Prof. A.S. Mishchenko and others (MSU), Acad. A.T.
Fomenko (MSU), Prof. E. Schrohe (Leibniz University of Hannover, Germany)]; at inter-
national conferences in Bialowieza (Poland), Voronezh, Saint-Petersburg, Tambov (Rus-
sia) and were taught many times at various scientific centers: Independent University of
Moscow, RUDN University, etc. We are grateful to all the participants of the seminars
and lectures for attention and constructive criticism during the talks. The authors are also
grateful to Vladimir Nazaikinskii and Pavel Sipailo.
During the preparation of this book, the authors were supported in part by the Si-
mons foundation and the Ministry of Education and Science of the Russian Federation
(agreement no. 02.a03.21.0008).
Preface v
1 Leray residues 1
1.1 A glimpse of one-dimensional residues . . . . . . . . . . . . . . . . . . . 1
1.2 First definition of Leray residues . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Second definition of Leray residues . . . . . . . . . . . . . . . . . . . . 3
1.4 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Leray exact sequences and residue theorem . . . . . . . . . . . . . . . . 5
1.6 Appendix. Some notions of (co)homology theory . . . . . . . . . . . . . 7
1.7 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2 Ramified integrals 11
2.1 Why do integrals ramify? . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 General theory. Landau manifolds . . . . . . . . . . . . . . . . . . . . . 14
2.3 Integrals over relative cycles . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Appendix. Differentiation of parametric integrals . . . . . . . . . . . . . 24
2.5 Appendix. Stratified sets . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.6 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
vii
viii Contents
Bibliography 131
Index 137
Chapter 1
Leray residues
res f (x)dx ∈ C
x=x0
and we set
res f (x)dx = a−1 . (1.2)
x=x0
These definitions of the residue of forms are invariant (in contrast to the definition
of residues of functions). More precisely, a direct computation shows that
both expressions (1.1) and (1.2) are invariant with respect to changes of variable!
Note also that the form f (x)dx is closed by Cauchy–Riemann equations.
Let us now translate (1.2) into the language of differential forms. To this end, we
substitute the Laurent expansion of f (x) in f (x)dx and represent this form as the sum
−2 ∞
an a−1 dx
f (x)dx = d ∑ (x − x0 ) n+1
+ + ∑ an (x − x0 )n dx (1.3)
n=−∞ n + 1 x − x0 n=0
of an exact form, a form with a first-order pole, and a holomorphic form. We can now
give the second definition of residue. Namely, it follows from (1.3) that the cohomology
class
[ f (x)dx] ∈ H 1 (C \ {x0 })
has the representative
dx
ϕ(x) + ψ(x)dx,
x − x0
where ϕ(x) and ψ(x) are regular functions at x = x0 . Then
Exercise 1.1. Show that (1.4) is invariant with respect to changes of variable.
Let us now turn to the multidimensional theory of residues.
X = {x ∈ X | s(x) = 0},
where ds = 0 on X. The function s(x) is called a (local) defining function of X. Under these
assumptions, a small tubular neighborhood1 of X can be represented as the total space of
a fibration with fiber equal to a small disc in C, hence, slightly simplifying the situation,
Leray residues can be thought of as residues in the direction normal to X, while we do
not take residues along the tangent directions. We now proceed to the precise definitions.
Let ω be a closed differential form of degree k that is smooth in a punctured neigh-
borhood of X, and consider a small tubular neighborhood of X. Denote the boundary of
this tubular neighborhood by SX (see Fig. 1.1). This boundary fibers over X with the fiber
S1 . Denote the projection for this bundle by
π : SX −→ X.
1 Recall
that a tubular neighborhood of radius r of a submanifold is the set of points of the ambient manifold,
whose distance to the submanifold is less then r.
1.3. Second definition of Leray residues 3
Definition 1.1. The Leray residue of the cohomology class [ω] ∈ H k (M \ X) of the form
ω is the cohomology class
1
Res[ω] = ω SX ∈ H k−1 (X), (1.5)
X 2πi S1
where
: Λk (SX) → Λk−1 (X)
S1
stands for the operator of integration along the fibers of the fibration SX → X (see the
Appendix to this chapter) oriented using the complex structure. Here and in what follows
Λk denotes the space of smooth k-forms on a manifold.
Clearly, in the one-dimensional case Definition 1.1 reduces to the classical definition
of residues using contour integration (1.1).
Let us also mention that the class (1.5) is well defined since integration along the
fiber (anti)commutes with the exterior differential.
I = (i1 , . . . , in ) is a multi-index, and the functions ωI (x) are holomorphic. Note that such
forms have no antiholomorphic differentials dx j . If the coefficients in (1.6) have singular-
ities, then we say that the form is analytic.
4 Chapter 1. Leray residues
where ψreg is taken from a representative of this class in terms of a form with a first-order
pole; see (1.7).
Remark 1.1. For forms with first-order poles (1.7), Definition 1.3 gives an explicit for-
mula for Leray residues. However, the computation of residues for forms with more com-
plicated singularities can be quite cumbersome. Nonetheless, for forms with poles2 of
order k ≥ 2 the residue can be calculated using the formula (see Pham [36])
ds θ ψ 1 dψ
ω= ψ + k−1 = d − + k−1 +θ
sk s (k − 1)sk−1 s k−1
and induction over k. Here the first summand on the right-hand side is exact, while the
second summand has a pole of order k − 1.
Exercise 1.2. Prove that Definitions 1.3 and 1.1 are equivalent using the Stokes formula.
1.4 Examples
1.
x1 dx1 ∧ dx2 ∧ dx3 x1 d(x1 + x2 x3 ) ∧ dx2 ∧ dx3
Res = Res
x1 +x2 x3 =0 x1 + x2 x3 x1 +x2 x3 =0 x1 + x2 x3
= −x2 x3 dx2 ∧ dx3 .
2 A form ω has a pole of order k on a submanifold {s(x) = 0}, if sk ω is holomorphic in a neighborhood of
this submanifold.
1.5. Leray exact sequences and residue theorem 5
2.
dζ 2 (x2 )2 dζ ∧ dx2
= Res x ∧ dx2 − Res
ζ =0 ζ ζ =0 ζ2
(x2 )2 dx2
= x2 dx2 − Res d − = x2 dx2 .
ζ =0 ζ
Here we set ζ = x1 + x2 .
3. (Poincaré 1887). Given any 1 ≤ j ≤ n, one has
j
f (x)dx1 ∧ · · · ∧ dxn dx1 ∧ · · · ∧ dx ∧ · · · ∧ dxn
Res = (−1) j+1 f (x)
s(x)=0 s(x) ∂s
∂xj
H ∗ (M) (1.9)
d
ω∗
y i∗
H ∗ (M \ X) / H ∗ (X)
δ ∗ =(2πi) Res
1 ds
ω ∗ϕ = − d χ ∧ π ∗ ϕ,
2πi s
where π : U → X stands for the projection of the tubular neighborhood on X, while χ is
a cut-off function equal to one on X and zero outside U. Finally, δ ∗ decreases degrees of
forms by one and takes a class to its Leray residue on X.
One can show that the triangle (1.9) is commutative.
6 Chapter 1. Leray residues
Leray sequence in homology. By duality (see the Appendix to this chapter) we also
have a commutative triangle of homology groups (with compact supports)
H∗ (M) (1.10)
9
i∗
ω
δ
$
H∗ (M \ X) o H∗ (X)
δ : H∗ (X) −→ H∗+1 (M \ X)
increases the dimension of cycles by one and takes a cycle on X to the pull-back of this
cycle on the circle bundle SX ⊂ M.
Let us give examples, where the Leray coboundary can be computed explicitly.
Example 1.1. Consider the embedding {0} ⊂ C of the origin. Then δ ({0}) is the homol-
ogy class of a circle around the origin.
Its real part X ∩ R2 is the circle defined by the same equation. The homology class of this
circle is denoted by
[X ∩ R2 ] ∈ H1 (X).
δ [X ∩ R2 ] ∈ H2 (C2 \ X)
Residues in relative theory. Suppose that we are also given a codimension one sub-
manifold S ⊂ M transversal to X. In this situation, we consider closed differential forms
on M that have no singularities on the complement of X, while their restriction to S is
zero. Recall that such forms define classes in relative de Rham cohomology H ∗ (M \ X, S).
Hereinafter, when we write a pair of spaces (A, B), we understand this pair as (A, A ∩ B).
In this relative theory the residue is a mapping
δ i∗ δ
· · · → H k−1 (X) −→ H k (M, X) −→ H k (M) −→ H k (X) −→ · · · , (1.12)
i∗ ∂ ∗ i
· · · → Hk (X) −→ Hk (M) −→ Hk (M, X) −→ Hk−1 (X) −→ ··· , (1.13)
where ∂ is the boundary mapping in homology (it takes each chain to its boundary).
8 Chapter 1. Leray residues
3. Duality between homology and cohomology. Integration of forms over chains de-
fines the de Rham pairing
H k (M) × Hk (M) −→ C
[ω] , [c] −→ ω
c
on a closed manifold M and also the pairing (defined by the same formula)
H k (M, X) × Hk (M, X) −→ C
4. Integration over the fiber. Let π : E → B be a locally trivial fiber bundle with com-
pact smooth base B, fiber F, and the total space E. Suppose that the fibration is oriented,
i.e., there is an orientation in the fibers that depends continuously on the point of the base.
In this case one can define the mapping of integration over the fiber denoted by
: Λk (E) −→ Λk−n (B), n = dim F. (1.14)
F
Recall the definition of (1.14). To this end we choose local coordinates on E and B such
that the projection π has the form
(x1 , . . . , xk , y1 , . . . , yn ) −→ (x1 , . . . , xk ).
F
∑ aI (x, y)dy1 ∧ · · · ∧ dyn ∧ dxI = ∑ Y
aI (x, y)dy1 · · · dyn ,
I I
where dxI = dxi1 ∧ dxi2 ∧ · · · ∧ dxik , and I = (i1 , i2 , . . . , in ) is a multi-index. Here we sup-
pose that y1 , . . . , yn is a positively oriented coordinate system on the fiber.
Integration over the fiber has the following properties (they can be checked by a di-
rect computation):
1) it is well defined (i.e., independent of the choice of coordinates);
2) for all ω1 ∈ Λ(E) and ω2 ∈ Λ(B) we have
(ω1 ∧ π ∗ ω2 ) = ω1 ∧ ω2 ;
F F
(This property means that integration over the fiber is a homomorphism of Λ(B)-
modules.)
3) given ω ∈ Λ(E), we have
d ω = (−1)n dω.
F F
1.6. Appendix. Some notions of (co)homology theory 9
5. Thom isomorphism and Thom class. Let E be an oriented vector bundle (with fiber
Rn ) over the base B.
Theorem 1.2 (Thom isomorphism). Integration over the fiber
: Hck (E) −→ H k−n (B) (1.15)
Rn
is an isomorphism for all k. The inverse mapping to (1.15) is called the Thom isomorphism
and is equal to the product with a special class
−1
TE = (1) ∈ Hcn (E), where 1 ∈ H 0 (B) is the generator,
Rn
called the Thom class. In other words, the group Hc∗ (E) is a free H ∗ (B)-module with one
generator equal to the Thom class.
Let us give an explicit formula for the Thom class of a one-dimensional complex
bundle E. Let us start with an example.
Example 1.3. Let C → pt be a one-dimensional complex line bundle over a point. Denote
the complex coordinate by x. Then we have Hc2 (C) C, and the Thom class is defined
by the 2-form
1 dx
d χ ,
2πi x
where χ ∈ C∞ (C) is a smooth function equal to zero at zero and equal to one at infinity.
A simple computation using Stokes’ formula shows that the integral of this 2-form is
equal to one. Hence, it can be taken as a representative of the Thom class.
Now let E be a trivial one-dimensional bundle E over an arbitrary base B. In this
case, the Thom class is defined by the same formula
1 ds
d χ ,
2πi s
3 It is also called the direct image mapping for the projection π.
10 Chapter 1. Leray residues
where χ ∈ C∞ (E) is a smooth function equal to zero in a neighborhood of the zero section
of E and equal to one at infinity, while s ∈ C∞ (E) is an arbitrary function, whose restric-
tion to the fiber of E is proportional to the complex coordinate on the fiber.4 A simple
computation shows that this 2-form is closed and its integral over the fiber is equal to one
identically; hence this form is indeed a representative of the Thom class for E.
Explicit expressions for differential forms representing Thom classes of nontrivial
bundles can be found in [5], Chapters 1 and 2; see also [30].
1.7 Remarks
Leray residues were defined above as cohomology classes, and the natural question arises:
is it possible to define the residue in terms of holomorphic forms representing this coho-
mology class? It turns out, that the answer is no: even though the residue is defined as
a cohomology class, it might be the case, that there is no holomorphic representative in
this class! A counterexample was constructed by Leray (see his book [27]). Thus, the
Leray residue of a holomorphic form can be a smooth (nonholomorphic) form. How-
ever, a holomorphic representative exists if we deal with Stein manifolds (e.g., see [1] for
details).
More results and applications of Leray residues can be found in (Leray [27], Chap-
ters 1-6; Pham [36], Chapter 3; Sternin and Shatalov [53]; Shabat [40]; see also the survey
by Dolbeaut [7] and the references therein).
Ramified integrals
where the contour of integration γ ⊂ C is a circle of radius 1/2 with center at x = 1 (see
Fig. 2.1).
Clearly, expression (2.1) determines f (t) as a holomorphic function in a neighbor-
hood of t = 1.1
The question then arises: how do we construct an analytic continuation of f (t) and
describe the singularities of this analytic continuation? Let us construct the analytic con-
tinuation of f (t), without explicitly computing the integral.2 Unfortunately, (2.1) does not
define f (t) for all t, since for some t zeroes of the denominator lie on the contour γ and
the integral diverges. The solution of this problem is obvious: we should simply move the
contour, or, more precisely, make the contour of integration depend on t. Indeed,
√ note that
(2.1) will not change, if√we take as the contour the circle with center at x = t and a small
radius such that x = − t lies outside this circle. Obviously, for t = 0 (i.e., when the roots
√
1 Indeed, the zeroes of the denominator are x = ± t, and therefore for t close to t = 1 the integrand has no
singularities on the contour of integration. Hence, the integral defines a holomorphic function of t by a standard
theorem of analysis on the differentiation of integrals, when the integrand√smoothly depends on the parameter.
2 This integral is easy to compute explicitly using the substitution x = tz.
do not coincide) we can choose a continuous family of such contours. For instance, we
can take the family (see Fig. 2.2)
√ √
γt = x ∈ C |x − t| = 12 | t| .
√
Roughly speaking, our contour γt is associated with the root x = t. However, this root
has ramification, when the parameter goes around t = 0. Hence, our contour also ramifies
when we go around this point. Thus, (2.1) also has ramification when we go around the
point t =√0, while its Riemannian surface coincides with the Riemannian surface of the
function t. Note, finally, that the singularity at t = 0 corresponds to the simple fact that
for this value of t the contour of integration gets pinched, that is, it is impossible to deform
it continuously in such a way that it avoids the singularities of the integrand.
Let us give a homological interpretation of the constructions above. For each t the
integrand in (2.1) defines a de Rham cohomology class
dx √
∈ H 1 Cx \ {± t} . (2.2)
x −t
2
Note that [γt ] is a ramified function of t. Finally, the analytic continuation of (2.1) is
defined as the pairing of the classes (2.2) and (2.3). Since (2.3) is ramified, the integral
(2.1) is a ramified function as well.
2. Integration over relative cycles. Consider the function f (t) defined by the paramet-
ric integral
f (t) = dx, (2.4)
γt
2.1. Why do integrals ramify? 13
√ √
where the chain γt ⊂ C is equal to the segment from − t to t. In this example the
integrand is nonsingular, but the chain has singularity.
Let us compute the ramification of (2.4) without computing the integral explicitly.
Obviously, the chain γt degenerates at t = 0 and has ramification of order two when we
go around the point t = 0. Thus, (2.4) has ramification of order two at t = 0.
Let us give a homological interpretation. The integrand in (2.4) defines a relative de
Rham cohomology class
√
[dx] ∈ H 1 Cx , {± t} . (2.5)
The chain γt defines a relative homology class
√
[γt ] ∈ H1 Cx , {± t} , (2.6)
which is a ramified function of t. Hence, the analytic continuation of (2.4) is given by the
pairing of classes (2.5) and (2.6).
Exercise 2.1. 1. Find singularities of the analytic continuation of the integral
dx
,
γ x2 + xt
where the contour γ goes around a) only one zero of the denominator; b) two zeroes of
the denominator; c) all the zeroes of the denominator.
3. Find singularities of the analytic continuation of the integrals
dx dy, dx dy dz.
x2 +y2 ≤t x2 +y2 +z2 ≤t
2. Construction of ramified
homology classes. Landau manifolds. Let us extend the
class [γ0 ] ∈ Hk X \ S(t0 ) up to a family of homology classes
[γt ] ∈ Hk X \ S(t) (2.8)
π : (X × T, S) −→ T
(x,t) −→ t
Theorem 2.1 (Thom isotopy theorem). The projection π is locally trivial on the comple-
ment T \ L provided that X is compact.
16 Chapter 2. Ramified integrals
Remark 2.1. The compactness condition in Thom’s theorem is necessary, as Fig. 2.5
shows. Here we consider the projection R2y,t → Rt , while S = {ty = 1}. Obviously, t = 0
does not lie on the Landau manifold, while at this point the projection is not locally trivial,
since points of S(t) go to infinity as t → 0. In applications, the compactness condition
means that we have to pass from the original problem on a noncompact space Cn to its
2.2. General theory. Landau manifolds 17
compactification CPn and, in particular, the Landau manifold also has contributions from
points at infinity.
Now we can construct the ramified homology class (2.8). Consider the bundle over
T with the fiber at t ∈ T equal to the homology group H∗ (X \ S(t)). Since π is locally
trivial over the complement T \ L of the Landau manifold, it follows that this bundle is
also locally trivial. This and the fact that H∗ (X \ S(t)) has discrete topology implies that
[γ0 ] extends uniquely up to a continuous family of homology classes
[γt ] ∈ Hk (X \ S(t))
π : (X × T, S) −→ T
defined over a Riemannian surface over T \ L, where L stands for the Landau manifold
for π.
4. Example. Let us find the singularities and the Riemannian surface of the integral
dx
f (t) = 2 −t
. (2.9)
γ(t) x
The set S = {x2 = t} ⊂ C2x,t is nonsingular. In this case the Landau manifold is just a one-
point space
L = {t = 0} ⊂ Ct ,
which is just the set of singularities of the projection of S to the t-axis. A ramified cycle
can be defined as √ 1√ √
γt = x |x − t| = | t| ⊂ Cx \ {± t}
2
√
and has Riemannian surface equal to that for the function t. √
Hence, the Riemannian surface for (2.9) coincides with that for the function t.
5. Stratified singularity sets. In general, the set S is stratified, i.e., it is a finite union
S= Aj
j
of smooth manifolds A j (called strata) such that the boundary A j \A j is a union of strata of
lower dimensions and the Whitney conditions are satisfied by any pair of adjacent strata
(for more details, see the Appendix to this chapter).
The results described above translate to the situation of stratified S with the follow-
ing natural changes in the definition of local triviality and Landau manifolds:
• in Definition 2.1 we need to suppose that homeomorphisms gt are stratified, i.e.,
they map strata to strata;
• in Definition 2.2 the Landau manifold for S should be defined as the union of Landau
manifolds of the strata A j .
Taking into account these changes, Thom’s Theorem 2.1 and Theorem 2.2 on the
singularities of parametric integrals remain valid.
L = {t = 0} ⊂ Ct .
and describe its singularity set. To this end, consider the projection
π : (X × T, S) −→ T (2.12)
of the pair (X × T, S). By Thom’s theorem this projection is locally trivial over T \ L,
where L stands for the Landau manifold. Hence, [γ0 ] extends to a ramified homology
class denoted by
[γt ] ∈ Hm X, S(t) ,
where t runs over a Riemannian surface T over T \ L.
Thus, the analytic continuation of (2.11) is defined by de Rham pairing of ω(t) and
γt , and is a holomorphic function on the Riemannian surface T , while the singularities of
the integral lie on the Landau manifold.
Example 2.3. Consider the integral
f (t) = dx ∧ dy, (2.13)
γt
where γt = (x, y) | x2 + y2 ≤ t . The analytic continuation of this relative cycle is equal
to √ √
γt = (u t, v t) ∈ C2 | u2 + v2 ≤ 1, u, v ∈ R .
In this case, the Landau manifold is L = {t = 0}. Indeed, we have a paraboloid S =
{x2 + y2 = t} (see Fig. 2.7). Clearly, the singularity of the projection of this surface to the
axis Rt is precisely the vertex x = y = t = 0. A direct computation shows that neither the
family of cycles
[γt ] ∈ H2 C2 , S(t) , where S(t) = (x, y) | x2 + y2 = t ,
√
Figure 2.9: Riemannian surface of 1 − t 2.
where Dt stands for the circular segment. The relative cycle in this integral defines the
class
[Dt ] ∈ H2 (C2 , Σ ∪ Lt ),
where Σ = {x2 + y2 = 1} ⊂ C2 .
Exercise 2.4. Show that [Dt ] has ramification of order two around the points
√ t = ±1,
while the Riemannian surface of f (t) is the same as that for the function 1 − t 2 , see
Fig. 2.9.
2. Integrals of holomorphic forms with zeroes and singularities. Let ω(t) be a closed
holomorphic form of degree m, which is equal to zero on a submanifold denoted by S(t)
and has singularities on a set denoted by Y (t). Such a form defines a relative cohomology
class
[ω(t)] ∈ H m X \Y (t), S(t) .
Suppose we are also given a homology class
γ0 ∈ Hm X \Y (t0 ), S(t0 ) .
π : (X × T, S ∪Y ) → T (2.15)
of the pair (X × T, S ∪ Y ). Let L be the Landau manifold for this projection. Then by
Thom theorem this projection is locally trivial over T \L. Hence, [γ0 ] extends to a ramified
homology class
[γt ] ∈ Hm X \Y (t), S(t) ,
Another random document with
no related content on Scribd:
artist, the decorator, the paper mills’ agent, and, last of all, the printer and the
binder. This was not the way the old-time printers had planned their books.
With all their mechanical limitations, they had followed architectural lines kept
consistent and harmonious because controlled by a single mind, while the
finished volume of the eighteen-nineties was a composite production of many
minds, with no architectural plan. No wonder that the volumes manufactured,
even in the most famous Presses, failed to compare with those produced in
Venice by Jenson and Aldus four centuries earlier!
When I succeeded John Wilson as head of the University Press in 1895, I
determined to carry out the resolution I had formed four years earlier, while
sitting in on the Eugene Field conference, of following the example of the
early master-printers so far as this could be done amidst modern conditions.
Some of my publisher friends were partially convinced by my contention that
if the printer properly fulfilled his function he must know how to express his
clients’ mental conception of the physical attributes of prospective volumes in
terms of type, paper, presswork, and binding better than they could do it
themselves. The Kelmscott publications, which appeared at this time, were of
great value in emphasizing my contention, for William Morris placed printing
back among the fine arts after it had lapsed into a trade.
I had no idea, when I presented my plan, of persuading my friends to
produce typographical monuments. No demand has ever existed for volumes
of this type adequate to the excessive cost involved by the perfection of
materials, the accuracy of editorial detail, the supreme excellence of
typography and presswork, and the glory of the binding. Sweynheim and
Pannartz, Gutenberg’s successors, were ruined by their experiments in Greek;
the Aldine Press in Venice was saved only by the intervention of Jean Grolier;
Henri Étienne was ruined by his famous Thesaurus, and Christophe Plantin
would have been bankrupted by his Polyglot Bible had he not retrieved his
fortunes by later and meaner publications. Nor was I unmindful of similar
examples that might have been cited from more modern efforts, made by
ambitious publishers and printers.
What I wanted to do was to build low-cost volumes upon the same
principles as de luxe editions, eliminating the expensive materials but retaining
the harmony and consistency that come from designing the book from an
architectural standpoint. It adds little to the expense to select a type that
properly expresses the thought which the author wishes to convey; or to have
the presses touch the letters into the paper in such a way as to become a part
of it, without that heavy impression which makes the reverse side appear like
an example of Braille; or to find a paper (even made by machine!) soft to the
feel and grateful to the eye, on which the page is placed with well-considered
margins; or to use illustrations or decorations, if warranted at all, in such a way
as to assist the imagination of the reader rather than to divert him from the
text; to plan a title page which, like the door to a house, invites the reader to
open it and proceed, its type lines carefully balanced with the blank; or to bind
(even in cloth!) with trig squares and with design or lettering in keeping with
the printing inside.
By degrees the publishers began to realize that this could be done, and
when once established, the idea of treating the making of books as a
manufacturing problem instead of as a series of contracts with different
concerns, no one of which knew what the others were doing, found favor. The
authors also preferred it, for their literary children now went forth to the
world in more becoming dress. Thus serving in the capacity of book architect
and typographical advisor, instead of merely as a contrasting printer, these
years have been lived in a veritable Kingdom of Books, in company with
interesting people,—authors and artists as well as publishers,—in a delightfully
intimate way because I have been permitted to be a part of the great
adventure.
During these years I have seen dramatic changes. Wages were somewhat
advanced between 1891 and the outbreak of the World War, but even at this
latter date the cost of manufacturing books was less than half of what it is now.
This is the great problem which publishers have to face today. When the cost
of everything doubled after the World War, the public accepted the necessity
of paying twice the price for a theater ticket as a matter of course; but when
the retail price of books was advanced in proportion to the cost of
manufacture, there was a great outcry among buyers that authors, publishers,
and booksellers were opportunists, demanding an unwarranted profit. As a
matter of fact, the novel which used to sell at $1.35 per copy should now sell
at $2.50 if the increased costs were properly apportioned. The publisher today
is forced to decline many promising first novels because the small margin of
profit demands a comparatively large first edition.
Unless a publisher can sell 5,000 copies as a minimum it is impossible for
him to make any profit upon a novel. Taking this as a basis, and a novel as
containing 320 pages, suppose we see how the $2.00 retail price distributes
itself. The cost of manufacture, including the typesetting, electrotype plates,
cover design, jacket, brass dies, presswork, paper, and binding, amounts to 42
cents per copy (in England, about 37 cents). The publisher’s cost of running
his office, which he calls “overhead,” is 36 cents per copy. The minimum
royalty received by an author is 10 per cent. of the retail price, which would
give him 20 cents. This makes a total cost of 98 cents a copy, without
advertising. But a book must be advertised.
Every fifty dollars spent in advertising on a five thousand edition adds a
cent to the publisher’s cost. The free copies distributed for press reviews
represent no trifling item. A thousand dollars is not a large amount to be spent
for advertising, and this means 20 cents a copy on a 5000 edition, making a
total cost of $1.18 per copy and reducing the publisher’s profit to 2 cents, since
he sells a two-dollar book to the retail bookseller for $1.20. The bookseller
figures that his cost of doing business is one-third the amount of his sales, or,
on a two-dollar book, 67 cents. This then shows a net profit to the retail
bookseller of 13 cents, to the publisher of 2 cents, and to the author of 20
cents a copy.
Beyond this, there is an additional expense to both bookseller and
publisher which the buyer of books is likely to overlook. It is impossible to
know just when the demand for a book will cease, and this means that the
publisher and the bookseller are frequently left with copies on hand which
have to be disposed of at a price below cost. This is an expense that has to be
included in the book business just as much as in handling fruit, flowers, or
other perishable goods.
When a publisher is able to figure on a large demand for the first edition,
he can cut down the cost of manufacture materially; but, on the other hand,
this is at least partially offset by the fact that authors whose books warrant
large first editions demand considerably more than 10 per cent. royalty, and
the advertising item on a big seller runs into large figures.
I wish I might say that I had seen a dramatic change in the methods
employed in the retail bookstores! There still exists, with a few notable
exceptions, the same lack of realization that familiarity with the goods one has
to sell is as necessary in merchandizing books as with any other commodity.
Salesmen in many otherwise well-organized retail bookstores are still painfully
ignorant of their proper functions and indifferent to the legitimate
requirements of their prospective customers.
Some years ago, when one of my novels was having its run, I happened to
be in New York at a time when a friend was sailing for Europe. He had
announced his intention of purchasing a copy of my book to read on the
steamer, and I asked him to permit me to send it to him with the author’s
compliments. Lest any reader be astonished to learn that an author ever buys a
copy of his own book, let me record the fact that except for the twelve which
form a part of his contract with the publisher, he pays cash for every copy he
gives away. Mark Twain dedicated the first edition of The Jumping Frog to “John
Smith.” In the second edition he omitted the dedication, explaining that in
dedicating the volume as he did, he had felt sure that at least all the John
Smiths would buy books. To his consternation he found that they all expected
complimentary copies, and he was hoist by his own petard!
With the idea of carrying out my promise to my friend, I stepped into one
of the largest bookstores in New York, and approached a clerk, asking him for
the book by title. My pride was somewhat hurt to find that even the name was
entirely unfamiliar to him. He ran over various volumes upon the counter, and
then turned to me, saying, “We don’t carry that book, but we have several
others here which I am sure you would like better.”
“Undoubtedly you have,” I agreed with him; “but that is beside the point. I
am the author of the book I asked for, and I wish to secure a copy to give to a
friend. I am surprised that a store like this does not carry it.”
Leaning nonchalantly on a large, circular pile of books near him, the clerk
took upon himself the education of the author.
“It would require a store much larger than this to carry every book that is
published, wouldn’t it?” he asked cheerfully. “Of course each author naturally
thinks his book should have the place of honor on the bookstalls, but we have
to be governed by the demand.”
It was humiliating to learn the real reason why this house failed to carry
my book. I had to say something to explain my presumption even in assuming
that I might find it there, so in my confusion I stammered,
“But I understood from the publishers that the book was selling very
well.”
“Oh, yes,” the clerk replied indulgently; “they have to say that to their
authors to keep them satisfied!”
With the matter thus definitely settled, nothing remained but to make my
escape as gracefully as circumstances would permit. As I started to leave, the
clerk resumed his standing position, and my eye happened to rest on the pile
of perhaps two hundred books upon which he had been half-reclining. The
jacket was strikingly familiar. Turning to the clerk I said severely,
“Would you mind glancing at that pile of books from which you have just
risen?”
“Oh!” he exclaimed, smiling and handing me a copy, “that is the very book
we were looking for, isn’t it?”
It seemed my opportunity to become the educator, and I seized it.
“Young man,” I said, “if you would discontinue the practice of letting my
books support you, and sell a few copies so that they might support me, it
would be a whole lot better for both of us.”
“Ha, ha!” he laughed, graciously pleased with my sally; “that’s a good line,
isn’t it? I really must read your book!”
The old-time publisher is passing, and the author is largely to blame. I have
seen the close association—in many cases the profound friendship—between
author and publisher broken by the commercialism fostered by some literary
agents and completed by competitive bids made by one publishing house to
beguile a popular author away from another. There was a time when a writer
was proud to be classified as a “Macmillan,” or a “Harper” author. He felt
himself a part of the publisher’s organization, and had no hesitation in taking
his literary problems to the editorial advisor of the house whose imprint
appeared upon the title pages of his volumes. A celebrated Boston authoress
once found herself absolutely at a standstill on a partially completed novel. She
confided her dilemma to her publisher, who immediately sent one of his
editorial staff to the rescue. They spent two weeks working together over the
manuscript, solved the problems, and the novel, when published, was the most
successful of the season.
Several publishers have acknowledged to me that in offering unusually
high royalties to authors they have no expectation of breaking even, but that to
have a popular title upon their list increases the sales of their entire line. The
publisher from whom the popular writer is filched has usually done his share
in helping him attain his popularity. The royalty he pays is a fair division of the
profits. He cannot, in justice to his other authors, pay him a further premium.
Ethics, perhaps, has no place in business, but the relation between author
and publisher seems to me to be beyond a business covenant. A publisher may
deliberately add an author to his list at a loss in order to accomplish a specific
purpose, but this practice cannot be continued indefinitely. A far-sighted
author will consider the matter seriously before he becomes an opportunist.
During the years that followed I served as his typographic mentor. He was
eager to try weird and ingenious experiments to bring out the various points of
his theories through unique typographical arrangement (see opp. page). It
required all my skill and diplomacy to convince him that type possessed rigid
limitations, and that to gain his emphasis he must adopt less complicated
methods. From this association we became the closest of friends, and
presuming upon this relation I used to banter him upon being so casual. His
copy was never ready when the compositors needed it; he was always late in
returning his proofs. The manufacture of a Fletcher book was a hectic
experience, yet no one ever seemed to take exceptions. This was characteristic
of the man. He moved and acted upon suddenly formed impulses, never
planning ahead yet always securing exactly what he wanted, and those
inconvenienced the most always seemed to enjoy it.
“I believe,” he used to say, “in hitching one’s wagon to a star, but I always
keep my bag packed and close at hand ready to change stars at a moment’s
notice. It is only by doing this that you can give things a chance to happen to
you.”
Among the volumes Fletcher had with him on board ship was one he had
purchased in Italy, printed in a type I did not recognize but which greatly
attracted me by its beauty. The book bore the imprint: Parma: Co’tipi Bodoniani.
Some weeks later, in a small, second-hand bookstore in Florence, I happened
upon a volume printed in the same type, which I purchased and took at once
to my friend, Doctor Guido Biagi, at the Laurenziana Library.
“The work of Giambattista Bodoni is not familiar to you?” he inquired in
surprise. “It is he who revived in Italy the glory of the Aldi. He and Firmin
Didot in Paris were the fathers of modern type design at the beginning of the
nineteenth century.”
“Is this type still in use?” I inquired.
“No,” Biagi answered. “When Bodoni died there was no one worthy to
continue its use, so his matrices and punches are kept intact, exactly as he left
them. They are on exhibition in the library at Parma, just as the old Plantin
relics are preserved in the museum at Antwerp.”
GIAMBATTISTA BODONI, 1740–1813
From Engraving at the Bibliothèque Nationale, Paris