224 Mr. E. Warren. 



As an example we may take the total breadth. Here the observed 

 range of variation is from 1370 to 1227 thousandths of the standard. 

 The unit of deviation employed is 0'004 of the carapace length ; thus, 

 with this unit the range of deviation in the total breadth is expressed 

 by the number 36. The crabs varied in absolute length of the 

 standard from 20'0 48'6 mm., and in the fourteen groups the 

 means are expressed in thousandths of carapace length, but the 

 errors of mean square and all the constants that will hereafter be 

 given are expressed in terms of the above unit of deviation. 



The length of the standard is taken as a criterion of age ; this 

 may, of course, be only roughly true, and it seems quite possible that 

 the somewhat large fluctuations which are observed in the means are 

 due to the groups being in reality slightly heterogeneous by including 

 crabs of somewhat different ages. 



On glancing down the means of the several dimensions, it will be 

 seen that a state of equilibrium is nowhere reached except perhaps 

 in the case of the total breadth; thus, throughout life the crab is 

 gradually changing its shape. This species, like the closely allied 

 Garcinus moenas, is probably sexually mature when only some 20 

 30 mm. long ;* hence, when all its organs are in a rapid state of 

 change the crab can propagate its species. Here is an argument, so 

 it would seem, against the transmission of acquired characters, for 

 otherwise the earlier broods would tend to have a somewhat different 

 shape to the later ones, and this is scarcely probable. To illustrate 

 this change in shape fig. 1 was prepared. The continuous outline 

 represents a crab with carapace length = 28'5 mm. ; the broken out- 

 line is the same crab when it has grown to 46'8 mm. ; the angle 

 CAC' opens out by about a degree. This, of course, is only true on 

 the supposition that natural selection has not occurred with respect to 

 these dimensions. The error of mean square of each group is given 

 in the columns to the right of the means. This constant we take as 

 a measure of the variability of an organ, and in no case is there a 

 distinct tendency for it to diminish as we pass down to the groups 

 containing the older crabs, while in the dentary margins there is an 

 obvious increase. Hence, at the present period in the life of the 

 species we have no evidence of selective destruction with regard to 

 the dimensions here discussed. It is possible, however, that all the 

 dimensions are really more variable as the crab grows older, but that 

 this greater variability is concealed by the action of natural selection 

 in all cases except in the dentary margins. 



We will now treat each organ separately. 



Total Breadth. On referring to the table, it will be seen that 

 groups 4 14 have means which remain * fairly steady, and show 



* This point is now being investigated. Female Carcinus occurs in berry when 

 only some 20 mm. long. 



