Hegel’s Science of Logic
Remark 2: The Purpose of the Differential Calculus Deduced from its Application
In the previous Remark we considered on the one hand the specific nature of the notion of the infinitesimal which is used in the differential calculus, and on the other the basis of its introduction into the calculus; both are abstract determinations and therefore in themselves also easy. The so-called application, however, presents greater difficulties, but also the more interesting side; the elements of this concrete side are to be the object of this Remark. The whole method of the differential calculus is complete in the proposition that dxn = nx(n - 1)dx, or (f(x + i) - fx)/i = P, that is, is equal to the coefficient of the first term of the binomial x + d, or x + 1, developed according to the powers of dx or i. There is no need to learn anything further: the development of the next forms, of the differential of a product, of an exponential magnitude and so on, follows mechanically; in little time, in half an hour perhaps — for with the finding of the differential the converse the finding of the original function from the differential, or integration, is also given — one can be in possession of the whole theory. What takes longer is simply the effort to understand, to make intelligible, how it is that, after having so easily accomplished the first stage of the task, the finding of the said differential, analytically, i.e. purely arithmetically, by the expansion of the function of the variable after this has received the form of a binomial by the addition of an increment; how it is that the second stage can be correct, namely the omission of all the terms except the first, of the series arising from the expansion. If all that were required were only this coefficient, then with its determination all that concerns the theory would, as we have said, be settled and done with in less than half an hour and the omission of the further terms of the series (with the determination of the first function, the determination of the second, third, etc., is also accomplished) far from causing any difficulty, would not come into question since they are completely irrelevant.
We may begin by remarking that the method of the differential calculus shows on the face of it that it was not invented and constructed for its own sake. Not only was it not invented for its own sake as another mode of analytical procedure; on the contrary, the arbitrary omission of terms arising from the expansion of a function is absolutely contrary to all mathematical principles, it being arbitrary in the sense that the whole of this development is nevertheless assumed to belong completely to the matter in hand, this being regarded as the difference between the developed function of a variable (after this has been given the form of a binomial) and the original function. The need for such a mode of procedure and the lack of any internal justification at once suggest that the origin and foundation must lie elsewhere. It happens in other sciences too, that what is placed at the beginning of a science as its elements and from which the principles of the science are supposed to be derived is not self-evident, and that it is rather in the sequel that the raison d'étre and proof of those elements is to be found. The course of events in the history of the differential calculus makes it plain that the matter had its origin mainly in the various so-called tangential methods, in what could be considered ingenious devices; it was only later that mathematicians reflected on the nature of the method after it had been extended to other objects, and reduced it to abstract formulae which they then also attempted to raise to the status of principle.
We have shown that the specific nature of the notion of the so-called infinitesimal is the qualitative nature of determinations of quantity which are related to each other primarily as quanta; to this was linked the empirical investigation aimed at demonstrating the presence of this specific nature in the existing descriptions and definitions of the infinitesimal in so far as this is taken as an infinitesimal difference and the like. This was done only in the interest of the abstract nature of the notion as such; the next question would be as to the nature of the transition from this to the mathematical formulation and application. To this end we must first pursue our examination of the theoretical side, the specific nature of the notion, which will not prove wholly unfruitful in itself; we must then consider the relation of the theoretical side to its application; and in both cases we must demonstrate, so far as it is relevant here, that the general conclusions are at the same time adequate to the purpose of the differential calculus and to the way in which the calculus brings about its results.
First, it is to be remembered that the mathematical form of the determinateness of the notion under discussion has already been stated in passing. The specifically qualitative character of quantity is first indicated in the quantitative relation as such; but it was already asserted in anticipation when demonstrating the so-called kinds of reckoning (see the relative Remark), that it is the relation of powers (still to be dealt with in its proper place) in which number, through the equating of the moments of its Notion, unit and amount, is posited as returned into itself, thereby receiving into itself the moment of infinity, of being-for-self, i.e. of being self-determined. Thus, as we have already said, the express qualitative nature of quantity is essentially connected with the forms of powers, and since the specific interest of the differential calculus is to operate with qualitative forms of magnitude, its own peculiar subject matter must be the treatment of forms of powers, and the whole range of problems, and their solutions, show that the interest lies solely in the treatment of determinations of powers as such.
This foundation is important and at once puts in the forefront something definite in place of the merely formal categories of variable, continuous or infinite magnitudes or even of functions generally; yet it is still too general, for other operations also have to do with determinations of powers. The raising to a power, extraction of a root, treatment of exponential magnitudes and logarithms, series, and equations of higher orders, the interest and concern of all these is solely with relations which are based on powers. Undoubtedly, these together constitute a system of the treatment of powers; but which of the various relations in which determinations of powers can be put is the peculiar interest and subject matter of the differential calculus, this is to be ascertained from the calculus itself, i.e. from its so-called applications. These are, in fact, the core of the whole business, the actual procedure in the mathematical solution of a certain group of problems; this procedure was earlier than the theory or general part and was later called application only with reference to the subsequently created theory, the aim of which was to draw up the general method of the procedure and, as well, to endow it with first principles, i.e. with a justification. We have shown in the preceding Remark the futility of the search for principles which would clarify the method as currently understood, principles which would really solve the contradiction revealed by the method instead of excusing it or covering it up merely by the insignificance of what is here to be omitted (but which really is required by mathematical procedure), or, by what amounts to the same thing, the possibility of infinite or arbitrary approximation and the like. If from the practical part of mathematics known as the differential calculus the general features of the method were to be abstracted in a manner different from that hitherto followed, then the said principles and the concern about them would also show themselves to be superfluous, just as they reveal themselves to be intrinsically false and permanently contradictory.
If we investigate this peculiarity by simply taking up what we find in this part of mathematics, we find as its subject matter:
(a) Equations in which any number of magnitudes (here we can simply confine ourselves to two) are combined into a qualitative whole in such a way that first, these equations have their determinateness in empirical magnitudes which are their fixed limits, and also in the kind of connection they have with these limits and with each other as is generally the case in an equation; but since there is only one equation for both magnitudes (similarly, relatively more equations for more magnitudes, but always fewer than the number of magnitudes), these equations belong to the class of indeterminate equations; and secondly, that one aspect of the determinateness of these magnitudes is that they are — or at least one of them is present in the equation in a higher power than the first.
Before proceeding further, there are one or two things to be noticed about this. The first is that the magnitudes, as described under the first of the above two headings, have simply and solely the character of variables such as occur in the problems of indeterminate analysis. Their value is undetermined, but if one of them does receive a completely determined value, i.e. a numerical value, from outside, then the other too, is determined, so that one is a function of the other. Therefore, in relation to the specific quantitative determinateness here in question, the categories of variable magnitudes, functions and the like are, as we have already said, merely formal, because they are still too general to contain that specific element on which the entire interest of the differential calculus is focused, or to permit of that element being explicated by analysis; they are in themselves simple, unimportant, easy determinations which are only made difficult by importing into them what they do not contain in order that this may then be derived from them — namely, the specific determination of the differential calculus. Then as regards the so-called constant, we can note that it is in the first place an indifferent empirical magnitude determining the variables only with respect to their empirical quantum as a limit of their minimum and maximum; but the nature of the connection between the constants and the variables is itself a significant factor in the nature of the particular function which these magnitudes are. Conversely, however, the constants themselves are also functions; in so far as a straight line, for example, has the meaning of being the parameter of a parabola, then this meaning is that it is the function y2/x2; and in the expansion of the binomial generally, the constant which is the coefficient of the first term of the development is the sum of the roots, the coefficient of the second is the sum of the products, in pairs, and so on; here, therefore, the constants are simply functions of the roots. Where, in the integral calculus, the constant is determined from the given formula, it is to that extent treated as a function of this. Further on we shall consider these coefficients in another character than that of functions, their meaning in the concrete object being the focus of the whole interest.
Now the difference between variables as considered in the differential calculus, and in their character as factors in indeterminate problems, must be seen to consist in what has been said, namely, that at least one of those variables (or even all of them), is found in a power higher than the first; and here again it is a matter of indifference whether they are all of the same higher power or are of unequal powers; their specific indeterminateness which they have here consists solely in this, that in such a relation of powers they are functions of one another. The alteration of variables is in this way qualitatively determined, and hence continuous, and this continuity, which again is itself only the purely formal category of an identity, of a determinateness which is preserved and remains self-same in the alteration, has here its determinate meaning, solely, that is, in the power-relation, which does not have a quantum for its exponent and which forms the non-quantitative, permanent determinateness of the ratio of the variables. For this reason it should be noted, in criticism of another formalism, that the first power is only a power in relation to higher powers; on its own, x is merely any indeterminate quantum. Thus there is no point in differentiating for their own sakes the equations y = ax + b (of the straight line), or s = ct (of the plain uniform velocity); if from y = ax, or even ax + b, we obtain a = dy/dx, or from s = ct, ds/dt = c, then a = y/x is equally the determination of the tangent, or s/t that of velocity simply as such. The latter is given the form of dy/dx in the context of what is said to be the development of the uniformly accelerated motion; but, as already remarked, the presence in the system of such a motion, of a moment of simple, merely uniform velocity, i.e. a velocity which is not determined by the higher power of one of the moments of the motion is itself an empty assumption based solely on the routine of the method. Since the method starts from the conception of the increment which the variable is supposed to acquire, then of course a variable which is only a function of the first power can also receive an increment; when now in order to find the differential we have to subtract the difference of the second equation thus produced from the given equation, the meaninglessness of the operation becomes apparent, for, as we have remarked, the equation for the so-called increments, both before and after the operation, is the same as for the variables themselves.
(b) What has been said determines the nature of the equation which is to be treated; we have now to indicate what is the interest on which the treatment of the equation is focused. This consideration can yield only known results, in a form found especially in Lagrange's version; but I have made the exposition completely elementary in order to eliminate the heterogeneous determinations associated with it. The basis of treatment of an equation of this kind shows itself to be this, that the power is taken as being within itself a relation or a system of relations. We said above that power is number which has reached the stage where it determines its own alteration, where its moments of unit and amount are identical — as previously shown, completely identical first in the square, formally (which makes no difference here) in higher powers. Now power is number (magnitude as the more general term may be preferred, but it is in itself always number), and hence a plurality, and also is represented as a sum; it can therefore be directly analysed into an arbitrary amount of numbers which have no further determination relatively to one another or to their sum, other than that together they are equal to the sum. But the power can also be split into a sum of differences which are determined by the form of the power. If the power is taken as a sum, then its radical number, the root, is also taken as a sum, and arbitrarily after manifold divisions, which manifoldness, however, is the indifferent, empirically quantitative element. The sum which the root is supposed to be, when reduced to its simple determinateness, i.e. to its genuine universality, is the binomial; all further increase in the number of terms is a mere repetition of the same determination and therefore meaningless.
[It springs solely from the formalism of that generality to which analysis perforce lays claim when, instead of taking (a + b)n for the expansion of powers, it gives the expression the form of (a + b + c + d...)n as happens too in many other cases; such a form is to be regarded as, so to speak, a mere affectation of a show of generality; the matter itself is exhausted in the binomial. It is through the expansion of the binomial that the law is found, and it is the law which is the genuine universality, not the external, mere repetition of the law which is all that is effected by this a + b + c + d ...]
The sole point of importance here is the qualitative determinateness of the terms resulting from the raising to a power of the root taken as a sum, and this determinateness lies solely in the alteration which the potentiation is. These terms, then, are wholly functions of potentiation and of the power. Now this representation of number as a sum of a plurality of terms which are functions of potentiation, and the finding of the form of such functions and also this sum from the plurality of those terms, in so far as this must depend solely on that form, this constitutes, as we know, the special theory of series. But in this connection it is essential to distinguish another object of interest, namely the relation of the fundamental magnitude itself (whose determinateness, since it is a complex, i.e. here an equation, includes within itself a power) to the functions of its potentiation. This relation, taken in complete abstraction from the previously mentioned interest of the sum, will show itself to be the sole standpoint yielded by the practical aspect of the science.
But first, another determination must be added to what has been said, or rather, one which is implied in it must be removed. It was said that the variable into the determination of which power enters is regarded as within itself a sum, in fact a system of terms in so far as these are functions of the potentiation, and that thus the root, too, is regarded as a sum and in the simply determined form of a binomial: xn = (y + z)n = (yn + ny(n-1)z + ... ). This exposition started from the sum as such for the expansion of the power, i.e. for obtaining the functions of its potentiation; but what is concerned here is not a sum as such, or the series arising from it; what is to be taken up from the sum is only the relation. The relation as such of the magnitudes is, on the one hand, all that remains after abstraction is made from the plus of a sum as such, and on the other hand, all that is needed for finding the functions produced by the expansion of the power.
But such relation is already determined by the fact that here the object is an equation, ym = axn, and so already a complex of several (variable) magnitudes which contains a power determination of them. In this complex, each of these variables is posited simply as in relation to the others with the meaning, one could say, of a plus implicit in it — as a function of the other variables; their character, that of being functions of one another, gives them this determination of a plus which, however, for that same reason, is wholly indeterminate — not an increase or an increment, or anything of that nature. Yet even this abstract point of view we could leave out of account; we can quite simply stop at the point where the variables in the equation having received the form of functions of one another, such functions containing a relation of powers, the functions of potentiation are then also compared with one another — these second functions being determined simply and solely by the potentiation itself. To treat an equation of the powers of its variables as a relation of the functions developed by potentiation can, in the first place, be said to be just a matter of choice or a possibility; the utility of such a transformation has to be indicated by some further purpose or use; and the sole reason for the transformation was its utility.
When we started above from the representation of these functions of potentiation of a variable which is taken as a sum complex within itself, this served only partly to indicate the nature of such functions, but partly also to show the way in which they are found.
What we have here then is the ordinary analytical development which for the purpose of the differential calculus is operated in this way, that an increment dx or i is given to the variable and then the power of the binomial is developed by the terms of the series belonging to it. But the so-called increment is supposed to be not a quantum but only a form, the whole value of which is that it assists the development; it is admitted — most categorically by Euler and Lagrange and in the previously mentioned conception of limit — that what is wanted is only the resulting power determinations of the variables, the so-called coefficients, namely, of the increment and its powers, according to which the series is ordered and to which the different coefficients belong. On this we could perhaps remark that since an increment (which has no quantum) is assumed only for the sake of the development, it would be most appropriate to take i (the one) for that purpose, for in the development this always occurs only as a factor; the factor one, therefore, fulfils the purpose, namely, that the increment is not to involve any quantitative determinateness or alteration; on the other hand, dx, which is burdened with the false idea of a quantitative difference, and other symbols like i with the mere show — pointless here — of generality, always have the appearance and pretension of a quantum and its powers; which pretension then involves the trouble that they must nevertheless be removed and left out. In order to retain the form of a series expanded on the basis of powers, the designations of the exponents as indices could equally well be attached to the one. But in any case, abstraction must be made from the series and from the determination of the coefficients according to their place in the series; the relation between all of them is the same; the second function is derived from the first in exactly the same manner as this is from the original function, and for the function counted as second, the first derived function is itself original. But the essential point of interest is not the series but simply and solely the determination of the power resulting from the expansion in its relation to the variable which for the power determination is immediate. It should not therefore be defined as the coefficient of the first term of the development, for it is first only in relation to the other terms following it in the series, and a power such as that of an increment, like the series itself, is here out of place; instead, the simple expression: derived function of a power, or as was said above: function of potentiation of a magnitude, would be preferable — the knowledge of the way in which the derivation is taken to be a development included within a power being presupposed.
Now if the strictly mathematical beginning in this part of analysis is nothing more than the finding of the function determined by the expansion of the power, the further question is what is to be done with the relation so obtained, where has it an application and use, or indeed, for what purpose are such functions sought. It is the finding of relations in a concrete subject matter which can be reduced to such a function that has given the differential calculus its great interest.
But as regards the applicableness of the relation, we need not wait for conclusions to be drawn from particular applications themselves, the answer follows directly and automatically from the nature of the matter which we have shown to consist in the form possessed by the moments of powers: namely, the expansion of the powers, which yields the functions of their potentiation, contains (ignoring any more precise determination) in the first place, simply the reduction of the magnitude to the next lower power. This operation is therefore applicable in the case of those objects in which there is also present such a difference of power determinations. Now if we reflect on the specific nature of space, we find that it contains the three dimensions which, in order to distinguish them from the abstract differences of height, length and breadth, we can call concrete — namely, line, surface and total space; and when they are taken in their simplest forms and with reference to self-determination and consequently to analytical dimensions, we have the straight line, plane surface and surface taken as a square, and the cube. The straight line has an empirical quantum, but with the plane there enters the qualitative element, the determination of power; further modifications, e.g. the fact that this also happens in the case of plane curves, we need not consider, for we are concerned primarily with the distinction in general. With this there arises, too, the need to pass from a higher power to a lower, and vice versa, when, for example, linear determinations are to be derived from given equations of the plane, or vice versa. Further, the motion in which we have to consider the quantitative relation of the space traversed to the time elapsed, manifests itself in the different determinations of a motion which is simply uniform, or uniformly accelerated, or alternately uniformly accelerated and uniformly retarded, and thus a self-returning motion; since these different kinds of motion are expressed in accordance with the quantitative relation of their moments, of space and time, their equations contain different determinations of powers, and when it is necessary to determine one kind of motion, or a spatial magnitude to which one kind of motion is linked, from another kind of motion, the operation also involves the passage from one power-function to another, either higher or lower. These two examples may suffice for the purpose for which they are cited.
The appearance of arbitrariness presented by the differential calculus in its applications would be clarified simply by an awareness of the nature of the spheres in which its application is permissible and of the peculiar need for and condition of this application. But now the further point of interest within these spheres themselves is to know between what parts of the subject matter of the mathematical problem such a relation occurs as is posited peculiarly by the differential calculus. First, it must be observed that there are two kinds of relation. The operation of depotentiating an equation considered according to the derivative functions of its variables, yields a result which, in itself, is no longer truly an equation but a relation; this relation is the subject matter of the differential calculus proper. This also gives us, secondly, the relation of the higher power form (the original equation) itself to the lower (the derivative). This second relation we must ignore for the time being; it will prove to be the special subject matter of the integral calculus.
Let us start by considering the first relation; for the determination of its moment (to be taken from the application, in which lies the interest of the operation) we shall take the simplest example from curves determined by an equation of the second degree. As we know, the relation of the co-ordinates is given directly by the equation in a power form. From the fundamental determination follow the determinations of the other straight lines connected with the co-ordinates, tangent, subtangent, normal, and so on.
But the equations between these lines and the co-ordinate are linear equations; the wholes with respect to which these lines are determined as parts, are right-angled triangles formed by straight lines. The transition from the original equation which contains the power form, to said linear equations, involves now the above-mentioned transition from the original function (which is an equation), to the derived function (which is a relation, a relation, that is, between certain lines contained in the curve). The problem consists in finding the connection between the relation of these lines and the equation of the curve.
It is not without interest, as regards the historical element, to remark this much, that the first discoverers could only record their findings in a wholly empirical manner without being able to account for the operation, which remained a completely external affair. It will be sufficient here to refer to Barrow, to him who was Newton's teacher. In his lect. Opt. et Geom., in which he treats problems of higher geometry according to the method of indivisibles, a method which, to begin with, is distinct from the characteristic feature of the differential calculus, he also puts on record' his procedure for determining tangents — 'because his friends urged him to do so'. To form a proper idea of how this procedure is formulated simply as an external rule, in the same style as the 'rule of three', or better still the so-called 'test by casting out nines', one must read Barrow's own exposition. He draws the tiny lines afterwards known as the increments in the characteristic triangle of a curve and then gives the instruction, in the form of a mere rule, to reject as superfluous the terms which, as a result of the expansion of the equations, appear as powers of the said increments or as products (etenim isti termini nihilum valebunt); similarly, the terms which contain only magnitudes to be found in the original equation are to be rejected (the subsequent subtraction of the original equation from that formed with the increments); and finally, for the increments of the ordinate and abscissa, the ordinate itself and the subtangent respectively are to be substituted. The procedure, if one may say so, can hardly be set forth in a more schoolmaster-like manner; the latter substitution is the assumption of the proportionality of the increments of the ordinate and the abscissa with the ordinate and the subtangent, an assumption on which is based the determination of the tangent in the ordinary differential method; in Barrow's rule this assumption appears in all its naive nakedness. A simple way of determining the subtangent was found; the artifices of Roberval and Fermat have a similar character. The method for finding maximal and minimal values from which Fermat started rests on the same basis and the same procedure. It was a mathematical craze of those times to find so-called methods, i.e. rules of that kind and to make a secret of them — which was not only easy, but in one respect even necessary, for the same reason that it was easy — namely, because the inventors had found only an empirical, external rule, not a method, i.e. nothing derived from established principles. Leibniz accepted such so-called methods from his contemporaries and so did Newton who got them directly from his teacher; by generalising their form and applicableness they opened up new paths for the sciences, but at the same time they also felt the need to wrest free the procedure from the shape of merely external rules and to try to procure for it the necessary justification.
If we analyse the method more closely, we find the genuine procedure to be as follows. Firstly the power forms (of the variables of course) contained in the equation are reduced to their first functions. But the value of the terms of the equation is thereby altered; there is now no longer an equation, but instead only a relation between the first function of the one variable and the first function of the other. Instead of px = y2 we have p : 2y, or instead of 2ax - x2 = y2, we have a - x : y, the relation which later came to be designated dy/dx. Now the equation represents a curve; but this relation, which is completely dependent on it and derived from it (above, according to a mere rule), is, on the contrary, a linear relation with which certain lines are in proportion: p : 2y or a - x : y are themselves relations of straight line of the curve, of the co-ordinates and parameters. But with all this, nothing is as yet known. The interest centres on finding that the derived relation applies to other lines connected with the curve, on finding the equality of two relations. And so there is, secondly, the question, which are the straight lines determined by the nature of the curve, standing in such a relation? But this is just what was already known: namely, that the relation so obtained is the relation of the ordinate to the subtangent. This the ancients had found in an ingenious geometrical manner; what the moderns have discovered is the empirical procedure of so preparing the equation of the curve that it yields that first relation of which it was already known that it is equal to a relation containing the line (here the subtangent) which is to be determined. Now on the one hand, this preparation of the equation — the differentiation — has been methodically conceived and executed; but on the other hand the imaginary increments of the co-ordinates and an imaginary characteristic triangle formed by them and by an equally imaginary increment of the tangent, have been invented in order that the proportionality of the ratio found by lowering the degree of the equation to the ratio formed by the ordinate and subtangent, may be represented, not as something only empirically accepted as an already familiar fact, but as something demonstrated. However, in the said form of rules, the already familiar fact reveals itself absolutely and unmistakably as the sole occasion and respective justification of the assumption of the characteristic triangle and the said proportionality.
Now Lagrange rejected this pretence and took the genuinely scientific course. We have to thank his method for bringing into prominence the real point of interest for it consists in separating the two transitions necessary for the solution of the problem and treating and proving each of them separately. One part of this solution (for the more detailed statement of the process we shall confine ourselves to the example of the elementary problem of finding the subtangent), the theoretical or general part, namely, the finding of the first function from the given equation of the is dealt with separately; the result is a linear relation, a curve, relation therefore of straight lines occurring in the system determined by the curve. The other part of the solution now is the finding of those lines in the curve which stand in this relation. Now this is effected in a direct manner i.e., without the characteristic triangle, which means that there is no assumption of infinitely small arcs, ordinates and abscissae, the last two being given the significance of dy and dx, that is, of being sides of that relation, and at the same time directly equating the infinitely small ordinate and abscissa with the ordinate and subtangent themselves. A line (and a point, too), is determined only in so far as it forms the side of a triangle and the determination of a point, too, falls only in such triangle. This, it may be mentioned in passing, is the fundamental proposition of analytical geometry from which are derived the co-ordinates of that science, just as (it is the same standpoint) in mechanics it gives rise to the parallelogram of forces, for which very reason the many efforts to find a proof of this latter are quite unnecessary. The subtangent, now, is made to be the side of a triangle whose other sides are the ordinate and the tangent connected to it. The equation of the latter, as a straight line, is p = aq (the determination does not require the additional term, + b which is added only on account of the fondness for generality); — the determination of the ratio p/q falls within a, the coefficient of q which is the respective first function (derivative) of the equation, but may simply be considered only as a = p/q being, as we have said, the essential determination of the straight line which is applied as tangent to the curve. But the first function (derivative) of the equation of the curve is equally the determination of a straight line; seeing then that the co-ordinate p of the first straight line and y, the co-ordinate of the curve, are assumed to be identical (so that the point at which the curve is touched by the first straight line assumed as tangent is also the starting point of the straight line determined by the first function of the curve), the problem is to show that this second straight line coincides with the first, i.e. is a tangent; or, algebraically expressed, that since y = fx and p = Fq, and it is assumed that y = p and hence that fx = Fq, therefore f'x = F'q. Now in order to show that the straight line applied as a tangent and the straight line determined by the first function of the equation coincide, and that therefore the latter is a tangent, Descartes has recourse to the increment i of the abscissa and to the increment of the ordinate determined by the expansion of the function. Thus here, too, the objectionable increment also makes its appearance; but its introduction for the purpose indicated and its role in the expansion of the function must be carefully distinguished from the previously mentioned employment of the increment in finding the differential equation and in the characteristic triangle. Its employment here is justified and necessary because it falls within the scope of geometry, the geometrical determination of a tangent as such implying that between it and the curve with which it has a point in common, no other straight line can be drawn which also passes through the said point. For, as thus determined, the quality of tangent or not-tangent is reduced to a quantitative difference, that line being the tangent of which simply greater smallness is predicated with respect to the determination in point. This seemingly only relative smallness contains no empirical element whatever, i.e. nothing dependent on a quantum as such; in virtue of the nature of the formula it is explicitly qualitative if the difference of the moments on which the magnitude to be compared depends is a difference of powers. Since this difference becomes that of i and i2 and i (which after all is meant to signify a number) is then to be conceived as a fraction, i2 is therefore in itself and explicitly smaller than i, so that the very conception of an arbitrary magnitude in connection with i is here superfluous and in fact out of place. For the same reason the demonstration of the greater smallness has nothing to do with an infinitesimal, which thus need not be brought in here at all.
I must also mention the tangential method of Descartes, if only for its beauty and its fame — well-deserved but nowadays mostly forgotten; it has, moreover, a bearing on the nature of equations and this, again, calls for a further remark. Descartes expounds this independent method, in which the required linear determination is likewise found from the same derivative function, in his geometry which has proved to be so fruitful in other respects tool; in it he has taught the great basis of the nature of equations and their geometrical construction, and also of the application of analysis, thereby greatly widened in its scope, to geometry. With him the problem took the form of drawing straight lines perpendicularly to given points on a curve as a method for determining the subtangent, etc. One can understand the satisfaction he felt at his discovery, which concerned an object of general scientific interest at that time and which is so purely geometrical and therefore was greatly superior to the mere rules of his rivals, referred to above. His words are as follows: 'J'ose dire que c'est ceci le probleme le plus utile et le plus general, non seulement que je sache, mais meme que j'aie l'amais desire de savoir en giometrie.' He bases his solution on the analytic equation of the right-angled triangle formed by the ordinate of the point on the curve to which the required straight line in the problem is to be drawn perpendicularly, by this same straight line (the normal), and thirdly, by that part of the axis which is cut off by the ordinate and the normal (the subnormal). Now from the known equation of a curve, the value of either the ordinate or the abscissa is substituted in the said triangle, the result being an equation of the second degree (and Descartes shows how even curves whose equations contain higher powers reduce to this); in this equation, only one of the variables occurs, namely, as a square and in the first degree — a quadratic equation which at first appears as a so-called impure equation. Descartes now makes the reflection that if the assumed point on the curve is imagined to be a point of intersection of the curve and of a circle, then this circle will also cut the curve in another point and we shall then get for the unequal xs thus produced, two equations with the same constants and of the same form, or else only one equation with unequal values of x. But the equation only becomes one for the one triangle in which the hypotenuse is perpendicular to the curve or is the normal, the case being conceived of in this way, that the two points of intersection of the curve and the circle are made to coincide and the circle is thus made to touch the curve. But in that case it is also true that the x or y of the quadratic equation no longer have unequal roots. Now since in a quadratic equation with two equal roots the coefficient of the term containing the unknown in the first power is twice the single root, we obtain an equation which yields the required determinations. This procedure must be regarded as the brilliant device of a genuinely analytical mind, in comparison with which the dogmatically assumed proportionality of the subtangent and the ordinate with the postulated infinitely small, so-called increments, of the abscissa and ordinate drops into the background.
The final equation obtained in this way, in which the coefficient of the second term of the quadratic equation is equated with the double root or unknown, is the same as that obtained by the method of the differential calculus. The differentiation of x2 - ax - b = 0 yields the new equation 2x - a = 0; or x3 - px - q = 0 gives 3x2 - p = 0. But it suggests itself here to remark that it is by no means self-evident that such a derivative equation is also correct. We have already pointed out that an equation with two variables (which, just because they are variables, do not lose their character of being unknown quantities) yields only a proportion; and for the simple reason stated, namely, that when the functions of potentiation are substituted for the powers themselves, the value of both terms of the equation is altered and it is not yet known whether an equation still exists between them with their values thus altered. All that the equation dy/dx = P expresses is that P is a ratio and no other real meaning can be ascribed to dyldx. But even so, we still do not know of this ratio = P, to what other ratio it is equal; and it is only such equation or proportionality which gives a value and meaning to it. We have already mentioned that this meaning, which was called the application, was taken from another source, empirically; similarly, in the case of the equations here under discussion which have been obtained by differentiation, it is from another source that we must know whether they have equal roots in order that we may learn whether the equation thus obtained is still correct. But this fact is not expressly brought to notice in the textbooks; it is disposed of, certainly, when an equation with one unknown, reduced to zero, is straightway equated with y, with the result, of course, that differentiation yields a dy/dx, i.e. only a ratio. The functional calculus, it is true, is supposed to deal with functions of potentiation and the differential calculus with differentials; but it by no means follows from this alone that the magnitudes from which the differentials or functions of potentiation are taken, are themselves supposed to be only functions of other magnitudes. Besides, in the theoretical part, in the instruction to derive the differentials, i.e. the functions of potentiation, there is no indication that the magnitudes which are to be subjected to such treatment are themselves supposed to be functions of other magnitudes.
Further, with regard to the omission of the constant when differentiating, we may draw attention to the fact that the omission has here the meaning that the constant plays no part in the determination of the roots if these are equal, the determination being exhausted by the coefficient of the second term of the equation: as in the example quoted from Descartes where the constant is itself the square of the roots, which therefore can be determined from the constant as well as from the coefficients — seeing that, like the coefficients, the constant is simply a function of the roots of the equation. In the usual exposition, the omission of the so-called constants (which are connected with the other terms only by plus and minus) results from the mere mechanism of the process of differentiation, in which to find the differential of a compound expression only the variables are given an increment, and the expression thereby formed is subtracted from the original expression. The meaning of the constants and of their omission, in what respect they are themselves functions and, as such, are or are not of service, are not discussed.
In connection with the omission of constants we may make a similar observation about the names of differentiation and integration as we did before about the expressions finite and infinite: that is, that the character of the operation in fact belies its name. To differentiate denotes that differences are posited, whereas the result of differentiating is, in fact, to reduce the dimensions of an equation, and to omit the constant is to remove from the equation an element in its determinateness. As we have remarked, the roots of the variables are made equal, and therefore their difference is cancelled. In integration, on the other hand, the constant must be added in again and although as a result the equation is integrated, it is so in the sense that the previously cancelled difference of the roots is restored, that is, what was posited as equal is differentiated again. The ordinary expression helps to obscure the essential nature of the matter and to set everything in a point of view which is not only subordinate but even alien to the main interest, the point of view, namely, of the infinitely small difference, the increment and the like, and also of the mere difference as such between the given and the derived function, without any indication of their specific, i.e. qualitative, difference.
Another important sphere in which the differential calculus is employed is mechanics. The meanings of the distinct power functions yielded by the elementary equations of its subject matter, motion, have already been mentioned in passing; at this point, I shall proceed to deal with them directly. The equation, i.e. the mathematical expression, for simply uniform motion, c = s/t or s = ct, in which the spaces traversed are proportional to the times elapsed in accordance with an empirical unit c (the magnitude of the velocity), offers no meaning for differentiation: the coefficient c is already completely determined and known, and no further expansion of powers is possible. We have already noticed how s = at2, the equation of the motion of a falling body, is analysed; the first term of the analysis, ds/dt = 2at is translated into language, and also into existence, in such a manner that it is supposed to be a factor in a sum (a conception we have long since abandoned), to be one part of the motion, which part moreover is attributed to the force of inertia, i.e. of a simply uniform motion, in such a manner that in infinitely small parts of time the motion is uniform, but in finite parts of time, i.e. in actually existent parts of time, it is non-uniform. Admittedly, fs = 2at; and the meaning of a and t themselves is known and so, too, the fact that the motion is determined as of uniform velocity; since a = s/t2, 2at is equal simply to 2s/t.
But knowing this we are not a whit wiser; it is only the erroneous assumption that 2at is a part of the motion regarded as a sum, that gives the false appearance of a physical proposition. The factor itself, a, the empirical unit — a simple quantum — is attributed to gravity; but if the category of 'force of gravity' is to be employed then it ought rather to be said that the whole, s = at2, is the effect, or, better, the law, of gravity. Similarly with the proposition derived from ds/dt = 2at, that if gravity ceased to act, the body, with the velocity reached at the end of its fall, would cover twice the distance it had traversed, in the same period of time as its fall. This also implies a metaphysics which is itself unsound: the end of the fall, or the end of a period of time in which the body has fallen, is itself still a period of time; if it were not, there would be assumed a state of rest and hence no velocity, for velocity can only be fixed in accordance with the space traversed in a period of time, not at its end. When, however, the differential calculus is applied without restriction in other departments of physics where there is no motion at all, as for example in the behaviour of light (apart from what is called its propagation in space) and in the application of quantitative determinations to colours, and the first function of a quadratic function here is also called a velocity, then this must be regarded as an even more illegitimate formalism of inventing an existence.
The motion represented by the equation s = at2 we find, says Lagrange, empirically in falling bodies; the next simplest motion would be that whose equation were s = ct3, but no such motion is found in Nature; we do not know what significance the coefficient c could have. Now though this is indeed the case, there is nevertheless a motion whose equation is s3 = at2 — Kepler's law of the motion of the bodies of the solar system; the significance 2at here of the first derived function 2at/3s2 and the further direct treatment of this equation by differentiation, the development of the laws and determinations of that absolute motion from this starting point, must indeed present an interesting problem in which analysis would display a brilliance most worthy of itself.
Thus the application of the differential calculus to the elementary equations of motion does not of itself offer any real interest; the formal interest comes from the general mechanism of the calculus. But another significance is acquired by the analysis of motion in connection with the determination of its trajectory; if this is a curve and its equation contains higher powers, then transitions are required from rectilinear functions, as functions of potentiation, to the powers themselves; and since the former have to be obtained from the original equation of motion containing the factor of time, this factor being eliminated, the powers must at the same time be reduced to the lower functions of development from which the said linear equations can be obtained. This aspect leads to the interesting feature of the other part of the differential calculus.
The aim of the foregoing has been to make prominent and to establish the simple, specific nature of the differential calculus and to demonstrate it in some elementary examples. Its nature has been found to consist in this, that from an equation of power functions the coefficient of the term of the expansion, the so-called first function, is obtained, and the relation which this first function represents is demonstrated in moments of the concrete subject matter, these moments being themselves determined by the equation so obtained between the two relations. We shall also briefly consider the principle of the integral calculus to see what light is thrown on its specific, concrete nature by the application of the principle. The view of the integral calculus has been simplified and more correctly determined merely by the fact that it is no longer taken to be a method of summation in which it appeared essentially connected with the form of series; the method was so named in contrast to differentiation where the increment counts as the essential element. The problem of this calculus is, in the first instance, like that of the differential calculus, theoretical or rather formal, but it is, as everyone knows, the converse of the latter. Here, the starting point is a function which is considered as deriv,ed, as the coefficient of the first term arising from the expansion of an equation as yet unknown, and the problem is to find the original power function from the derivative; what would be regarded in the natural order of the expansion as the original function is here derived, and the function previously regarded as derived is here the given, or simply original, function. Now the formal part of this operation seems to have been accomplished already in the differential calculus in which the transition and the relation of the original to the derived function in general has been established. Although in doing this it is necessary in many cases to have recourse to the form of series simply in order to obtain the function which is to be the starting point and also to effect the transition from it to the original function, it is important to remember that this form as such has nothing directly to do with the peculiar principle of integration.
The other part of the problem of the calculus appears in connection with its formal operation, namely the application of the latter. But this now is itself the problem: namely, to find the meaning in the above-mentioned sense, possessed by the original function of the given function (regarded as first) of a particular subject matter; it might seem that this doctrine, too, was in principle already finally settled in the differential calculus; but a further circumstance is involved which prevents the matter from being so simple. In the differential calculus, namely, it was found that the linear relation is obtained from the first function of the equation of a curve, so that it is also known that the integration of this relation gives the equation of the curve in the relation of abscissa and ordinate; or, if the equation for the area enclosed by the curve were given, then we should be supposed to know already from the differential calculus that the meaning of the first function of such equation would be that it represented the ordinate as a function of the abscissa, and therefore the equation of the curve.
The problem now is to determine which of the moments determining the subject matter is given in the equation itself; for the analytical treatment can only start from what is given and then pass on to the other moments of the subject matter. What is given is, for example, not the equation of an area enclosed by the curve, nor, say, of the figure resulting from its rotation; nor again of an arc of the curve, but only the relation of the abscissa and ordinate in the equation of the curve itself. Consequently, the transitions from those determinations to this equation itself cannot yet be dealt with in the differential calculus; the finding of these relations is reserved for the integral calculus.
But further, it has been shown that the differentiation of an equation of several variables yields the derived function or differential coefficient, not as an equation but only in the form of a ratio; the problem is then to find in the moments of the given subject matter a second ratio that is equal to this first ratio which is the derived function. By contrast, the object of the integral calculus is the relation itself of the original to the derived function, which latter is here supposed to be given; so that the problem concerns the meaning to be assigned to the sought-for original function in the subject matter of the given first derived function; or rather, since this meaning, for example, the area enclosed by a curve or the rectification of a curve represented as a straight line, already finds expression in the statement of the problem, to show that an original function has that meaning, and which is the moment of the subject matter which must be assumed for this purpose as the initial function of the derived function.
Now the usual method makes the matter easy for itself by using the idea of the infinitesimal difference; for the quadrature of curves, an infinitely small rectangle, a product of the ordinate into the element, i.e. the infinitesimal bit of the abscissa, is taken for the trapezium one of whose sides is the infinitely small arc opposite to the infinitesimal bit of the abscissa; the product is now integrated in the sense that the integral is the sum of the infinitely many trapezia or the area to be determined — namely, the finite magnitude of this element of the area. Similarly, from an infinitely small element of the arc and the corresponding ordinate and abscissa, the ordinary method forms a right-angled triangle in which the square of the arc element is supposed to be equal to the sum of the squares of the two other infinitely small elements, the integration of this giving the length of the arc itself as a finite quantity.
This procedure rests on the general discovery on which this field of analysis is based, in this instance, namely, that the quadrated curve, or the rectified arc, stands to a certain function given by the equation of the curve, in the relation of the so-called original function to its derivative. The aim of the integral calculus is this: when a certain part of a mathematical object (e.g. of a curve) is assumed to be the derived function, which other part of the object is expressed by the corresponding original function? It is known that when the function of the ordinate given by the equation of the curve is taken as the derived function, the corresponding original function gives the quantitative expression for the area of the curve cut off by this ordinate; and, when a certain tangential determination is identified with the derived function, the corresponding original function expresses the length of the arc belonging to this tangential determination, and so on. But the method which employs the infinitesimal, and operates with it mechanically, simply makes use of the discovery that these relations — the one of an original function to its derivative and the other of the magnitudes of two parts or elements of the mathematical object — form a proportion, and spares itself the trouble of demonstrating the truth of what it simply presupposes as a fact. The singular merit here of mathematical acumen is to have found out from results already known elsewhere, that certain specific aspects of a mathematical object stand in the relationship to each other of the original to the derived function.
Of these two functions it is the derived function or, as it has been defined, the function of potentiation, which here in the integral calculus is given relatively to the original, which has first to be found by integration. But the derived function is not directly given, nor is it at once evident which part or element of the mathematical object is to be correlated with the derived function in order that by reducing this to the original function there may be found that other part or element, whose magnitude is required to be determined. The usual method, as we have said, begins by representing certain parts of the object as infinitely small in the form of derived functions determinable from the originally given equation of the object simply by differentiation (like the infinitely small abscissae and ordinates in connection with the rectification of a curve); the parts selected are those which can be brought into a certain relation (one established in elementary mathematics) with the subject matter of the problem (in the given example, with the arc) this, too, being represented as infinitely small, and from this relation the magnitude required to be known can be found from the known magnitude of the parts originally taken. Thus, in connection with the rectification of curves, the three infinitely small elements mentioned are connected in the equation of the right-angled triangle, while for the quadrature of curves, seeing that area is taken arithmetically to be simply the product of lines, the ordinate and the infinitely small abscissa are connected in the form of a product. The transition from such so-called elements of the area, the arc, etc., to the magnitude of the total area or the whole arc itself, passes merely for the ascent from the infinite expression to the finite expression, or to the sum of the infinitely many elements of which the required magnitude is supposed to consist.
It is therefore merely superficial to say that the integral calculus is simply the converse, although in general the more difficult, problem of the differential calculus; the real interest of the integral calculus concerns almost exclusively the relation between the original and the derived function in the concrete subject matter.
Even in this part of the calculus, Lagrange did not smooth over the difficulties of its problems simply by making those direct assumptions. It will help to elucidate the nature of the matter in hand if here, too, we indicate the details of his method in one or two examples. The declared object of his method is, precisely, to provide an independent proof of the fact that between particular elements of a mathematical whole, for example, of a curve, there exists a relation of the original to the derived function. Now this proof cannot be effected in a direct manner because of the nature of the relation itself in this domain; in the mathematical object this relation connects terms which are qualitative distinct, namely, curves with straight lines, linear dimensions and their functions with plane or surface dimensions and their functions, so that the required determination can only be taken as the mean between a greater and a less. Consequently, there spontaneously enters again the form of an increment with a plus and minus and the energetic 'developpons' is here in place; but we have already pointed out that here the increments have only an arithmetical, finite meaning. From the development of the condition that the required magnitude is greater than the one easily determinable limit and smaller than the other, it is then deduced that, e.g. the function of the ordinate is the derived, first function of the function of the area.
Lagrange's exposition of the rectification of curves in which he starts from the principle of Archimedes is interesting because it provides an insight into the translation of the Archimedean method into the principle of modern analysis, thus enabling us to see into the inner, true meaning of the procedure which in the other method is carried out mechanically. The mode of procedure is necessarily analogous to the one just indicated. The principle of Archimedes, that the arc of a curve is greater than its chord and smaller than the sum of the two tangents drawn through the end points of the arc and contained between these points and the point of intersection of the tangents, gives no direct equation, but simply postulates an endless alternation between terms determined as too great or too small, the successive terms always being still too great or too small but within ever narrower limits of inaccuracy; its translation into the modern analytical form, however, takes the form of finding an expression which is per se a simple fundamental equation. Now whereas the formalism of the infinitesimal directly presents us with the equation dz2 = dx2 + dy2, Lagrange's exposition, starting from the basis indicated, demonstrates that the length of the arc is the original function to a derived function whose characteristic term is itself a function coming from the relation of a derived function to the original function of the ordinate.
Because in Archimedes' method, as well as later in Kepler's treatment of stereometric objects, the idea of the infinitesimal occurs, this has often been cited as an authority for the employment of this idea in the differential calculus, although what is peculiar and distinctive in it has not been brought specifically to notice. The infinitesimal signifies, strictly, the negation of quantum as quantum, that is, of a so-called finite expression, of the completed determinateness possessed by quantum as such. Similarly, in the subsequent celebrated methods of Valerius and Cavalieri, among others, which are based on the treatment of the relations of geometrical objects, the fundamental principle is that the quantum as such of the objects concerned, which are primarily considered only in their constituent relations, is for this purpose to be left out of account, the objects thus being taken as non-quantitative.
However, in these methods the affirmative aspect as such which is veiled by the merely negative determination fails to be recognised or brought to notice — that aspect namely which above presented itself abstractly, as the qualitative determinateness of quantity, and more precisely, as lying in the relation of powers; and also, since this relation itself embraces a number of more precisely determined relations such as that of a power and the function of its development; these also, in turn, are supposed to be based on and derived from the general and negative determination of the same infinitesimal. In the exposition of Lagrange just noticed, the specific affirmative aspect which is implied in Archimedes' method of developing the problem is brought to notice with the result that the procedure which is burdened with an unlimited progression is given its proper limit. The greatness of the modern invention per se and its capacity to solve previously intractable problems and to treat in a simple manner those previously soluble, is to be ascribed solely to the discovery of the relation of the original to the so-called derived functions and of those parts of a mathematical whole which stand in such a relation.
What has been said may suffice to signalise that distinctive relation of magnitudes which is the subject matter of the particular kind of calculus under discussion. It was possible to confine our exposition to simple problems and the methods of solving them, it would neither have been expedient as regards the determination of the Notion, which determination is here our sole concern, not would it have lain in the author's power to have reviewed the entire compass of the so-called application of the differential and integral calculus, and by reference of all the respective problems and their solutions to what we have demonstrated to be the principle of the calculus, to have carried out completely the induction that the application is based upon this principle. But sufficient evidence has been produced to show that just as each particular mode of calculation has as its subject matter a specific determinateness or relation of magnitude, such relation constituting addition, multiplication, the raising to powers and extraction of roots, and operations with logarithms and series, and so on, so too has the differential and integral calculus; the subject matter proper to this calculus might be most appropriately named the relation between a power function and the function of its expansion or potentiation, because this is what is most readily suggested by an insight into the nature of the subject matter.
Logarithms, circular functions and series are of course also employed in the calculus, especially for the purpose of making expressions more amenable for the operations necessary for deriving the original function from the functions of expansion; but they are only used in the same way that the other forms of calculation such as addition, etc., are also used in the calculus. The differential and integral calculus has, indeed, a more particular interest in common with the form of series namely, to determine those functions of expansion which in the series are called coefficients of the terms; but whereas the calculus is concerned only with the relation of the original function to the coefficient of the first term of its expansion, the series aims at exhibiting in the form of a sum, groups of the terms arranged according to powers which have these coefficients. The infinite which is associated with infinite series, the indeterminate expression of the negative of quantum in general, has nothing in common with the affirmative determination belonging to the infinite of this calculus. Similarly, the infinitesimal in the shape of the increment, by means of which the expansion is given the form of a series, is only an external means for the expansion, and the sole meaning of its so-called infinity is to have no other meaning beyond its significance as such means; the series, which in fact is not what is wanted, produces an excess, the elimination of which causes the unnecessary trouble. The method of Lagrange, who preferred to use the form of series again, is also burdened with this difficulty; although it is through his method, in what is called the application, that what is truly characteristic of the calculus is brought to notice, for, without forcing the forms of dx, dy and so on, into the objects, it is directly demonstrated to which part of the object the determinateness of the derived function (function of expansion) belongs; and thus it is evident that the matter in hand here is not the form of series.
[In the critique quoted above are to be found interesting views of a profound scholar in this science, Herr Spehr; they are quoted from his Neue Prinzipien des Fluentenkalkuls, Brunswick, 1826, and concern a factor which has materially contributed to what is obscure and unscientific in the differential calculus and they agree with what we have said about the general character of the theory of this calculus. 'Purely arithmetical investigations,' he says, 'admittedly those which have a primary bearing on the differential calculus, have not been separated from the differential calculus proper, and in fact, as with Lagrange, have even been taken to be the calculus itself whilst this latter was regarded as only the application of them. These arithmetical investigations include the rules of differentiation, the derivation of Taylor's theorem, etc., and even the various methods of integration. But the case is quite the reverse, for it is precisely those applications which form the subject matter of the differential calculus proper, all those arithmetical developments and operations being presupposed by the calculus from analysis.' We have shown how, with Lagrange, it is just the separation of the so-called application from the procedure of the general part which starts from series, which serves to bring to notice the characteristic subject matter of the differential calculus. It is strange, however, that the author, who realises that it is just these applications which form the subject matter of the differential calculus proper, should get involved in the formal metaphysics (adduced in that work) of continuous magnitude, becoming, flow, etc., and should want to add even fresh ballast to the old; these determinations are formal, in that they are only general categories which do not indicate just what is the specific nature of the subject matter, this having to be learned and abstracted from the concrete theory, that is, the applications.]
Remark 3: Qualitative Determinateness of Magnitude - next section
Hegel-by-HyperText Home Page @ marxists.org