448 



Mr. W. F. R. Weldon. The Variations [Apr. 17, 



curve and the curve of error, the ordinates of the observed curve 

 will be compared with those of the curve of error at certain fixed 

 grades. 



This may be most conveniently done by considering the ordinates 

 of the cnrve of error at those grades as multiples of the " probable 

 error " of the curve. 



The grades chosen, together with the ordinates of a curve of error, 

 expressed in terms of its probable error, at those grades, are as 

 follows : — 



Table I. — Ordinates of a Curve of Error, in Terms of Q. 



Grrade. 



Ordinate = Qx 



Grrade. 



Ordinate = Q, x 



5° 



+ 2-44 



60° 



-0-38 



10 



+ 1-90 



70 



-0-78 



20 



+ 1-25 



75 



-1 -00 



25 



+ 1-00 



80 



-1-25 



30 



+ 0-V8 



90 



-1 -90 



40 



+ 0-38 . 



95 



-2-44 



50 



o-oo 







It will be seen that, in order to compare a curve constructed from 

 a number of observations with a curve of error, the following process 

 is performed : the ordinates at the selected grades are determined, 

 and the observed valne of M is subtracted from each of these. The 

 remainders, divided by i(Qi — Q 3 ), should give the coefiicients of Q 

 which appear in the above table. 



Such a comparison will now be made between the curve of error 

 and the curves obtained from the measurements. 



The organs measured are four : the total length of the carapace ; 

 the distance from the posterior margin of the carapace to the front 

 of the median spine; the length of the sixth abdominal tergum; and 

 the length of the telson. The parts are shown in the accompanying 

 woodcufc. 



Fig. 1. 



