(10) If p, r, be the perpendicular on the tangent and p2 the radius vector at any point of a curve, then will be the perpendicular on the tangent at the corresponding point of the curve which is the locus of the extremity of p. Let x, y, be the co-ordinates of the first curve, a, ß, of the second; then p being the perpendicular on the tangent, its equation is since a, ß, are the co-ordinates of the extremity of p. But the line being a tangent, this equation will hold when we put x + dx and y + dx for x and y; we then have Now if V = 0 be the equation connecting a and ß, that is to say, the equation to the locus of the extremity of p, and if P be the perpendicular on the tangent of that curve, Now differentiating (1) considering x, y, a, ß, as variables, and paying attention to (2), we have (x − 2a) da + (y – 2 ß) dẞ = 0. (4) λ(3) (4)=0 gives, on equating to zero the coefficients 2 (a2 + ẞ2) − (ax + ẞy) P = [x2 + y2 + 4 {a2 + ß2 − (ax + ẞy)} ]' (11) To find the least polygon of a given number of sides which will circumscribe a given oval figure. Let AB, BC, CD, (fig. 26) be consecutive sides of the polygon. Produce AB, DC to meet in E, which take as origin, the axes being EA, ED. Then the position of BC must be such as to make BEC a maximum. Now calling as before the intercepts of the tangent ≈。, Yo, x and y being the co-ordinates of the point of contact P. The area BEC = y sin E, therefore is to be a maximum, (neglecting the negative sign). Differentiate with respect to æ, The last factor alone gives a solution. From it we have That is, EMEB = MB, and hence also CP = PB, As the same con or CB is bisected at the point of contact. dition holds for every side of the polygon, it follows that, when the polygon circumscribing an oval is a minimum, each side is bisected at the point of contact. Hence we see that of all the parallelograms which circumscribe an ellipse, those are least which have their sides parallel to conjugate diameters. R (12) The degree of a curve being n, there cannot be more than n(n - 1) tangents drawn to it from one point. Let u = c (1) be the equation to the curve, then the equation to the tangent is and the condition that this tangent shall pass through a The equations (1) and (2) being combined together will give the values of x and y at the points of contact; and as both equations are of n dimensions in a and y, (since u is du dx + y du dy of n dimensions), it would appear that the re sulting equation is of the degree n2, and therefore that there are as many tangents passing through the point. But the degree of the equation can always be reduced; for we may combine (2) with any multiple of (1), and the result of the elimination between the new equation and either of the others will still give us the co-ordinates of the point of contact. Multiply (1) by n and subtract it from (2), then we have For Now by a property of homogeneous functions, if v be homogeneous of n dimensions in x and y, This then will be true of the terms of n dimensions in u, and they will therefore disappear from the second side of the equation (3), which will thus be reduced to (n − 1) dimensions, since du du and are only of that degree. Hence the comdx dy bination of (1) with (3) will rise only to the degree n (n − 1), which therefore represents the greatest number of tangents which can be drawn from a given point to a curve of n dimensions. Waring had fixed the limit at n2, as it at first sight appears to be; the preceding process of reduction is due to Bobillier, Annales de Gergonne, Vol. xix. p. 106. It is to be observed that though n (n − 1) is the greatest number of tangents which can be drawn, it seldom reaches that limit, since the final equation generally involves impossible roots which refer to tangents drawn to the branches of the curve which do not lie in the plane xy. Since n (n-1) is essentially even, it may happen that for certain positions of the point all the roots are impossible; a result which is geometrically apparent, inasmuch as from the interior of an oval curve, such as the ellipse, no tangents can be drawn to the part of the curve which lies in the plane of ay. Asymptotes. As an asymptote is a line which, intersecting the axes at a finite distance from the origin, is a tangent to the curve at an infinite distance, it appears that if x, or y remain finite when a or y are infinite, their values will determine the position of the asymptote. A more convenient method however is that first given by Stirling, in his Lineæ Tertii ordinis Newtonianæ, p. 48. If y = f(x) be the equation to the curve, and if we can expand f(x) in descending powers of a, so that y = amxm + Am−1xm−1 + &c. + α ̧x + α ̧ + + a -2 + &c.; then when = ∞, the terms involving negative powers of a vanish, and the equation to the curve coincides with that to another curve the equation to which is y = ɑ„xTM +ɑm_1xTM-1 + &c. + a1x + α。⋅ amxm This then is the general equation to a curvilinear asymptote, the nature of which will depend on the degree of the highest power of a which is involved in it. The most important case is that in which the equation is reduced to y = а1x + ao, that is, in which the asymptote is a straight line. This method fails when the asymptote is parallel to the axis of y, as in that case the coefficient of a would be infinite: but asymptotes of this kind are visible by a simple inspection of the equation to the curve when it is put under the form y = f(x). For the value of y being infinite for the abscissa corresponding to the asymptote, we have only to find what values of a will make f(x) = ∞, or to make the denominator of f(x) vanish, since no finite value of x in the numerator can make f(x)= ∞. These values of a being found, the ordinates drawn through them are asymptotes to the curve. (13) Let the equation to the curve be But from the equation to the curve, 3 (3 x3) = 3 ax2, therefore + 1 = 1 when x and y are infinite. |