EXCLUSIVELY MADE USE OF.

LATE FELLOW OF CAIUS COLLEGE, CAMBRIDGE.

PRINTED AT THE UNIVERSITY PRESS;

PUBLISHED BY J. & J. J. DEIGHTON, CAMBRIDGE;

AND

JOHN W. PARKER, LONDON.

**PREFACE**

The method of Limits is generally allowed to be the best and most natural basis upon which to found the principles of the Differential Calculus; in the following pages this method is exclusively adopted, no use whatever being made of series in the demonstration of fundamental propositions. The following is an outline of the work, which is by no means offered to the reader as a complete treatise on the subject, but merely as an exposition of its more prominent and useful principles.

In Chap. I, certain terms, afterwards to be used, are defined and explained. In Chap. II the nature of a Limiting Value is fully set forth, and the important distinction (which ought never to be overlooked) between an actual and a limiting value is pointed out and illustrated by examples. Chap. III contains a set of Lemmas, which are necessary in order to render the use made of limiting values in the Differential Calculus perfectly legitimate; and here I have endeavoured to confine myself to what seems really essential. In Chap. IV certain important limiting values are obtained. Chap. V contains the Rules for Differentiation, in the demonstration of which Lagrange's functional notation is employed, as being the simplest to begin with. In Chap. VI the Differential notation of Leibniz is explained, ^{dy}/_{dx} is defined as the quote of the differentials *dy* and *dx*, which however are not supposed to be infinitesimals, but simply two arbitrary quantities in a certain ratio. In the case of partial differential coefficients, some modification of the common differential notation ^{du}/_{dx}, ^{du}/_{dy}, is clearly necessary: I have employed the suffix notation *d*_{x}*u*, *d*_{y}*u*, as being frequently employed, though not exactly in this manner. I should have much preferred the notation *d*_{x}*u*, *d*_{y}*u* to denote partial differentials, and ^{dxu}/_{dx}, ^{dyu}/_{dy} to denote partial differential coefficients. Chap. VII relates to successive differentiation, and the change of the independent variable. Chap. VIII contains certain very important Lemmas upon which the use and application of the Differential Calculus in a great measure depends. Chap. IX contains the theory of Series, based upon one of the preceding Lemmas, without assuming that *f* (*x* + *h*) can be developed in the form

*A*+

*Bh*

^{a}+

*Ch*

^{b}+ &

*c*. ...

It was my intention to have added a few more chapters, and among the rest, one on the origin and progress of the Differential Calculus, and another on the Infinitesimal method; but from various circumstances I found it impossible to send the work to the press at the time originally promised to my bookseller, without omitting these concluding chapters. I mention this to account for the absence of allusions to the History of the Differential Calculus, which were all reserved for the final chapter, and the small number of Examples in the Appendix.

Professor Peacock's excellent collection of Examples, which have been of such service to the Mathematical Student, is now out of print; but Mr Gregory's work lately published will supply its place, which contains, not only a great number of well-selected and valuable examples, but also many important explanations and theorems not to be met with in any elementary treatise. In a subject of so much importance as the present, the student ought not to confine his attention to one book or system: for a very valuable treatise on this subject he is referred to that published by the Society for the Diffusion of Useful Knowledge.

In the general plan of this work, and in several particulars, I have deviated from some of the methods often made use of, partly in attempting to put the subject in a simpler and clearer point of view, and partly in avoiding certain steps of reasoning which appear to be defective. One of these is the fallacy of establishing premises on a certain implied condition, and drawing a conclusion from them by a direct violation of that condition. An example of this is to be found in a proof often given of the principle of indeterminate coefficients, in which the factor *x* is divided out of the equation

*Bx*+

*Cx*

^{2}+

*Dx*

^{3}+ &

*c*. ... = 0,

*B*+

*Cx*+

*Dx*

^{2}+ &

*c*. ... = 0;

The assumption, that f(x + h) can be expanded in a series of the form *A* + *Bh*^{a} + *Ch*^{b} + &*c*. ... seems to me to be a serious defect in the common method of establishing Taylor's Series, and thereupon the principles of the Differential Calculus. This assumption is usually justified by arguing, that if we find definite values for *A*, *B*, *C*, &*c*. it shows that the assumption is correct. Now this argument may be stated thus:

*f*(

*x*+

*h*) =

*A*+

*Bh*

^{a}+

*Ch*

^{b}+ &

*c*. be true, then

*A*,

*B*,

*C*, &

*c*., must have definite values. But we can in general obtain definite values for

*A*,

*B*,

*C*, &

*c*.... (e.g. by the method of indeterminate coefficients.) Therefore the assumption is true."

*A*,

*B*,

*C*, &

*c*."

Cambridge,

October, 1842