216 University of California Publications in Zoology \|Vou.9 
when subjected to a chemical environment radically different 
from anything either it or any of its ancestors were ever sub- 
jected to, is of the greatest significance to philosophical biology, 
but could be discovered by no other means than experiment. 
This sort of discovery is not restricted to laboratory experimen- 
tation though undoubtedly some of the most startling, most 
definite results have been and will continue to be reached by 
this method. The transference of a plant from a desert interior 
to a seaside is obviously much the same kind of experiment as 
that of the transference of an aquatic plant from water holding 
one chemical substance in solution to water holding another 
substance. 
3y whatever method either the form or the behavior of an 
organism be modified through experiment, a sine qua non to 
sound interpretation of the new form or behavior is extensive 
and exact comparison with the form or behavior of the same 
organism under natural conditions; and such comparison is im- 
possible without extensive and exact description of the organisms 
both as they oceur in nature and under the new conditions. So 
we are forced back even from a consideration of this higher use 
of experimentation to the demand for the most extensive and 
the most. exact studies possible of organisms not only as they 
have developed and lived their lives in nature, but while they 
are developing and living their lives in nature. 
These general and special considerations as to method taken 
together with a consideration of the problems now in hand and 
of the resources of the station, indicate very definitely not only 
what ought to be done in the near future but also what may be 
aimed at with prospects of a good measure of success. 
As to field work the foremost desideratum is that it should 
be made more continuous. This means practically that the 
“* Agassiz”’ 
should be kept in constant commission though not 
necessarily at sea all the time. The main point is not so much 
to increase the volume and scope of the operations, though this 
would result, as to complete what is already in progress. Great 
gaps exist in the data so far secured, and consequently most of 
the conclusions as to daily, yearly, and vertical distribution and 
