622 THE AMERICAN NATURALIST. (VoL. XXXIV. 
of development), the variability depends upon numerous 
minute positively and negatively acting causes, a part of which 
only affects, in each case, the simple individuals of form unit 
in different combinations. Then the relative frequency of the. 
individual variants in a form unit corresponds to the possibility 
of these combinations. Therefore, the causes of variation are to 
be thought of as essentially different from the conditions deter- 
mining the average value of a character in the total form unit. 
On the other hand, I assume a relation between the morpho- 
logical and physiological peculiarity of the different characters 
and the causes of their individual variation, so that the single 
characters are only able to react on a part of the sum total of 
the latter ; therefore, in allied species homologous characters 
ought to show similar indices of variability, but not necessarily 
equal average values. 3 
In 1892 Weldon investigated the numbers of dorsal and ven- 
tral spines of the rostrum in 915 individuals of Palemonetes 
varians Leach from Saltram Park, near Plymouth, England. 
When I learned from Dr. Davenport that these numbers are 
markedly higher in the closely allied, if not identical, species, 
Palemonetes vulgaris Herbst, at Cold Spring Harbor, I took 
up the investigation of 1050 individuals of the latter form in 
order to try, in at least one case, which opinion held true. If 
there existed any relation between average and variability, 
the latter ought to be sensibly higher in P. vulgaris than 
in P. varians. If I was right, the variability of the number 
of rostral spines, in spite of the higher average in P. vulgaris, 
ought to be about the same in both species. 
I shall follow Weldon and deal with the males and females 
together, for among the 1050 individuals caught in the seine, 
only 92 were males. The males are much smaller than the 
females. The empirical results obtained from P. vulgaris are 
given in a table of combinations, those of 7. varians (for 
comparison) in series of variations. 
Explanation of letters used in the tables : 
J, empirical, y, theoretical frequencies ; 7, total number of 
individuals investigated ; A, error between empirical and theo- 
retical series of variation in percentages of n ; A, arithmetic 
