8 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I49 



sampling when sediment type permits. Walton (1955) discussed the 

 advantages of using equal wet volumes rather than dry sediment 

 weights in foramini feral ecology. 



A sample is assumed to be representative of both the distribution 

 and abundance of the foramini fers at the sampling site (station) as 

 well as of the total area the sample represents. Phleger (1952) has 

 indicated that in the Gulf of Maine the foraminiferal samples are 

 representative of the total area a sample represents because the distri- 

 bution of species is not haphazard, has localized centers or highs, 

 and decreases away from these highs in an orderly manner. Walton 

 (1955) discussed the same problem in his study of Todos Santos Bay, 

 Calif. The percentage distributions of the living species in the Bay 

 indicated the highest rate of variation at depths of less than 50 

 fathoms. In deeper water the amount of fluctuation diminishes. 

 Because of the stability of the percentage distribution of species in 

 deeper areas, Walton concluded that his sampling grid was giving 

 an adequate representation of foraminiferal distribution. 



In L.I.S. the percentage distribution of foraminiferal species is 

 meaningful and repeatable. This suggests that the samples from the 

 Sound are representative of the foraminiferal distribution in the area. 

 In order to test the reliability of a sample at a station at a particular 

 time, 12 pairs of samples were taken at various locations. Each 

 member of a sample pair was taken within minutes of the other, by 

 the same method, and at the same location as far as conditions would 

 permit. Theoretically, each pair should be identical. Sample pairs 

 14-14' and 125-125' were grabs, all the rest were cores. In the pair 

 14-14' the dead population was estimated. 



Statistical Significance of Species Proportions 



The species proportions in the sample pairs will now be compared. 

 The data are viewed most conveniently by arranging them in a con- 

 tingency table (see table 1, page 65). A qualitative approach would 

 be to compare visually the number of individuals in each species of a 

 sample pair and decide arbitrarily whether or not the species propor- 

 tions are similar. If the species proportions differ widely, then the 

 samples are considered not homogeneous. A more quantitative 

 approach is to choose a statistic which will test for homogeneity of 

 sample pairs. In the present study the statistic chosen was chi- 

 square. Because one of the assumptions on which this statistic is 

 based is violated if the frequency in a given category is too small, only 

 the three most abundant species were used in making the calculations. 



