PRIMARY PRODUCTION 535 



relations between the different factors involved, they are a neces- 

 sary link between the data and that other set of idealized pictures, 

 laboratory experiments. The two are complementary and together 

 can allow some insight into the causal relations operating in the sea. 



However, if there is to be any hope of showing whether the 

 hypothetical relations are actually occurring in the sea, one must 

 make a very careful choice of areas to work in. It is no longer true 

 that any part of the sea can be made to yield information that is 

 useful from this point of view. For a randomly chosen area, the 

 hydrographic structure during any period of time is likely to be 

 very complicated and difficult to understand. As a consequence 

 the biological cycles at different times or places will not be suffi- 

 ciently similar to be considered together nor sufficiently different 

 to be easily separable. 



Thus the amount of information that can be acquired about the 

 production cycle will depend on the hydrographic information 

 available and on the way in which this information can be ex- 

 pressed as a fairly simple picture. We need to use aieas that 

 approximate to two-layered seas: regions that are laterally homo- 

 geneous and without the troublesome occasional incursions of dif- 

 ferent water masses; sharp hydrographic edges; or such special 

 cases as the "dome" regions studied by Holmes which may cor- 

 respond to diffusion of nutrients from a centre. 



Yet, although the choice of areas is based in hydrographic con- 

 siderations, the distribution of sampling is not. The classical con- 

 cept of lines or grids of stations at widely spaced intervals of 

 time and distance, which has been the main method of revealing 

 hydrographic patterns, is not necessarily suited to plankton prob- 

 lems. In the extreme case of patchiness in temperate waters, very 

 detailed sampling in a very homogeneous hydrographic area may 

 be needed to show up the biological patterns in this process. 



Similarly, the usefulness of data on primary production for 

 fishery problems is still limited. One is likely to get the sort of 

 randomness already described, if the problems are chosen on the 

 basis of the economic importance of the fisheries rather than on 

 the probable simplicity of the main features of the environment. 

 In this way small or isolated fisheries may be the best testing 



