f 



The chi-square test v/as applied to the length frequencies shovm in 

 Table 8 grouped into classes and using the mean of the two series as the 

 expected frequency » The resulting chi square of 60 62 indicated the 

 difference betvfeen expected frequency and that observed in each of the 

 series would be exceeded 76 times out of one hundred in a homogeneous 

 population o Excluding those clams below 31 nrni, in length changes chi 

 square to 2,755 and the resulting probability to 91%. Exclusion of these 

 small clams is Justified on the basis that their enumeration is subject 

 to greater error because of their size and because counts vrere based on 

 l/U square foot subsaraples taken from the regular five-square-foot bottom 

 samples o 



The size composition of the population in the control area can there- 

 fore be reliably estimated from series of SO clamshell bucket samples. 



Length frequencies for the bullraked and dredged areas were not 

 separated into series of 50 samples each and therefore cannot be analyzed 

 by the chi-square testo If these areas can be considered analagous to 

 the controlj the same limits of accuracy can be applied. The length fre- 

 quencies in Figures 9^ 11 and 13 were based on 100 samples, so it is 

 likely that they are reliable estimates of the size composition of the 

 population on the test plot in 19^0 o Length frequencies shown by the 

 solid line in Figures 8^ 10 and 12 are based on 28 clamsViell bucket 

 samples, so it is likely that these provide less accurate estimates of 

 the size composition of the clam population,. 



Ii3 U9837 



