METEORS OF AUGUST AND NOVEMBER. 135 



cannot be identical. The differences may be embraced in two classes, those 

 which increase with the time, and those which arise from discrepancies of the 

 elements. Denoting the former by A, and the latter by d, and their joint 

 effect by 8, we have in the interval {f — t), 



S^ = ^'-^ = ^ + dx + A{x-X) 



(18) Syi = y}' — yi = y! + d7/ + A{y — Y) 

 8^ = ^' -^=^ + dz + A{z-Z) 



for the variations of ^, yj, and ^. Hence, in estimating the variations Sy, (5^, 

 and S(3, some allowance must be made for the quantities dy, d/l, and d/?, 

 arising from discrepancies of the true elements of the meteors seen at the dates 

 t, f, &c. Professor Erman's formulae, on the contrary, proceed upon the pre- 

 sumption of d^ = — AX, &c. Now, as d:r, + Az, &c., cannot, even in the 

 thickest flocks of meteors, vanish entirely their aggregate effect in a finite in- 

 terval of a few hours, may be such as to preponderate over that of A ( — X), and 

 in this manner the variation ^;i, in a few hours, may come out positive, as re- 

 ported by Professor Forshey, August 9th, 1840. Also, on ordinary nights, if 

 a convergent point is found to prevail, we should have 



(19) 6^ = ^' ^^ = ^ + {dx + Ax)-'AX=8z — AX 



and so on for 8yj and 8^; now if during an interval ^' — ^ of several nights we 

 find by observation, with Mr. Fitch and Mr. Herrick, 82, = A (L + 180,) 

 8^ = — A -B = 0, we are led to the inference that 8y = A (— G), and d ^ = — ' 

 AX, &c. And that a compensation has taken place among the true velocities 

 and directions of the meteors seen near each date, so that the convergent point 

 has corresponded, in position and variations, with those of the antipode of the 

 observer's actual direction. 



On the occasion of great displays like that of November, 1833, when tele- 

 scopes were directed to the radiant point, and its altitude was measured with 

 a sextant, it is probable that d x, dy, and d z were very small ; in such a case 

 a precise measure of the position of this point in the heavens might possibly, 

 by giving the value of 82. and 5/? in a finite interval, enable us to determine y 

 from the terms dx, dy, and d z, of which it would be a function. I do not, 

 however, think such a precision can ever be obtained. 



There is another point of view in which the knowledge of the values of d;?;, 



