222 POPULATIONS OF THE SEA 



Gulland (1955) has made the case for small samples in the study 

 of commercial fish populations. 



Hedgpeth (1957b) wrote that "the most significant aspect of 

 the intertidal populations of sandy beaches, at least in temperate 

 latitudes, is its fluctuation." I would have thought that this was 

 true of all marine life and that all the major problems facing 

 ecologists are concerned with the mechanisms of stability and 

 change in the sea, problems which can be solved only by a study 

 of fluctuations whether they be short-term, long-term, seasonal, 

 annual, spatial, or temporal. 



It is obvious that results must be expressed in quantitative 

 terms so that changes in abundance and distribution can be 

 measured and compared. This usually means making a count of 

 some kind. "Quantitative," however, is not a synonym for "ac- 

 curate" in the sense of "precise." Whatever the temporal or 

 spatial frequency of sampling, the interpretation of the results 

 will involve processes of interpolation and extrapolation which 

 may sometimes be closer to acts of faith than to scientific methods. 

 The largest samplers take extremely small subsamples of most 

 marine populations and, from what we already know of variation 

 in the sea, the estimation of population parameters from even 

 the most unbiased of samples is a hazardous business. 



With these points in mind careful thought should be given to 

 the desired accuracy of the techniques. It is necessary to ensure 

 that the time-consuming practice of accurate counting does not 

 add useless refinement at the expense of frequency of sampling. 

 In the Plankton Recorder Survey we have compromised between 

 counting and estimating the abundance of organisms; for instance, 

 the number of a species in a subsample is expressed as lying be- 

 tween 5 and 11, or 26 and 50, or 251 and 500. Such categories for 

 estimates of abundance must be chosen with care so that they do 

 not inhibit the subsequent mathematical treatment of the results. 



A further problem in the design of surveys of distribution and 

 abundance is concerned with the techniques of identification of 

 organisms. Many useful contributions have been made from a 

 knowledge of the total weight or volume of organic matter, or 



