THE CAUSES OF VARIATION. 231 



from their nature not capable of verification by experi- 

 ment ; anci (ii.) theories which regard them as due to 

 the action of environment, and which are therefore to 

 some extent capable of being put to actual experimental 

 test. 



In the present paper the writer proposes to give a short 

 account of some experiments he has recently been making 

 for the purpose of determining the exact amount of varia- 

 tion which is produced in the size of an organism by definite 

 changes in the conditions of environment. Similar experi- 

 ments have of course been made before, but as a rule the 

 changes in the environment have been made of such a kind 

 and such a degree as do not occur in a state of nature, and, 

 moreover, it is but seldom that the actual numerical effect 

 upon the size of the organism has been carefully measured. 

 In the present experiments the changes of environment 

 were, in the majority of cases, only such as might occur 

 under natural conditions, and the effect produced was 

 accurately determined by measuring large numbers of the 

 organisms allowed to develop under the varying conditions, 

 and calculating mean values. For such experiments as 

 these, it is essential to work with an organism that can be 

 readily obtained in large numbers, and which can be easily 

 kept under artificial conditions. The larvae or plutei of sea- 

 urchins answer to such requirements most admirably, and 

 they were in fact the organisms chosen. Thus these larvae 

 can always be obtained in innumerable quantities by the 

 simple process of artificially fertilising ova obtained by 

 shaking pieces of the ovaries of the sea-urchins in water, by 

 adding other water containing spermatozoa obtained in a 

 similar manner by shaking pieces of the testes. In a day 

 or two these fertilised ova, which are transferred to glass 

 jars holding 2 or 3^ litres of water, develop into free swim- 

 ming larvae. These continue to grow in size up to a certain 

 point. In order to measure their size, they are killed by 

 adding '25 per cent, of corrosive sublimate to the water, pre- 

 serving them in alcohol and, after mounting them in gly- 

 cerine, measuring them under the microscope with a 



micrometer eyepiece. These larvae are peculiarly easy to 



16 



