There wo.s really very little difference in the accuracy of the 

 two techniques (Appendix F, Table 1). By the volumetric technique 

 the calculated totals differed from the actual totals (niamerical 

 counts) by amounts varying from 0.0 to 10. I4 percent. The mean 

 percentage error was plus 3.2 percent. By the gravimetric tech- 

 nique, deviations from actual totals varied from 0.0 to 9-7 per- 

 cent ivith a mean percentage error of plus 2.5 percent. The mean 

 percentage errors of the two techniques are based on an algebraic 

 average of the individual percentage errors. Although in most 

 cases this statistic might produce an inaccurate result, it is 

 felt that in this instance its application is justified. A mean 

 error based on the algebraic sum of deviations expressed in numbers 

 of eggs allows individual specimens (such as a small one of low egg 

 count) to influence the results unduly, i.e., a female, for example, 

 containing only 2^,000 eggs and for tiiiich there is a plus or minus 

 1,000 egg deviation between calculated and actual totals has a 

 large percentage error. This latter statistic is a very real 

 measiure of the efficiency of the calculating technique in that 

 particular case. However, an equal numerical deviation in a larger 

 an.d more productive specimen results in a much lower percentage 

 error. It follows, then, that if errors expressed in numbers of 

 eggs tend to remain more or less constant (as they may if there is 

 some small bias in the technique), too many small test specimens 

 or too many large test specimens in a series mil undoubtedly 

 render too high or too low a mean error. 



The preponderance of positive errors in calculating total egg 

 production suggests some small bias in the techniques or procedures 

 used. It is felt that the mean percentage error, as computed, pro- 

 vides 'the best measure of any bias, if it exists, resulting from 

 some defect in the procedure. The mean percentage errors in this 

 study are so small, however, that they may very well fall within 

 the limits of chance occurrence. Consequently, the application of 

 any small correction factor to the calculated data on egg produc- 

 tion is not suggested. 



The gravimetric method was utilized in calculating total egg 

 production for the remaining 60 specimens because, apparently, the 

 smallest error was involved in this technique and because the 

 laboratory procedure is more rapid by this method. Data on siae^ 

 weight; and egg production were tabulated in order of increasing 

 octal length of specimens (Appendix F, Tkble 2). Wiere there wa» 

 more than one specimen of a given length, these Trere listed in 

 order of jjicreasing egg production. These data have been plotted 

 upon two graphs to illustrate the relationship betvreen egg produc- 

 tion and total length (Figure 26) and betvreen egg production and 

 weight (Figure 27). 



The first property evident in these figures is that the number 

 of eggs produced by the sea lamprey varies greatly at any particu- 

 lar length or iveight. Mean egg production was computed by one-inch 

 size groups and by 50-gram weight groups and these values were 

 plotted upon the respective graphs. The curves appearing in 



- 79 - 



