71 



From them horizontals are drawn to cut the curve, and from 

 the points of intersection perpendiculars are drawn to cut the 

 horizontal axis. The latter points are the two quartiles. 



It must be noted that the degree of accuracy of such 

 determinations of the modes, points of inflexion, or measures 

 of dispersion depends upon good draughtsmanship. This is 

 not difficult to attain. There will be some personal differences 

 in the results,* but we submit that these are usually smaller 

 than any differences that ought to affect the conclusions that are 

 to be made. These conclusions are to have certain probabili- 

 ties : for instance, we lay it down that it is to be 2 to 1 that 

 the fish caught on a certain area, in a certain month, and with 

 a 6-inch mesh trawl-net, are (say) between n and m cms. in 

 length. Then we find n and m., or we wish to find what fraction 

 of the whole catch of fish are between n and m cms. long in the 

 same area and in the same circumstances. Then n and m 

 being postulated we find the corresponding probability. 

 Extensions will readily suggest themselves. 



Characteristic Lengths of the Plaice Caught During 

 THE Pre-War Period, 1908-1913. 



We now give a series of tables, 3 to 13, which summarise 

 the results of the trawling experiments made during the six 

 years 1908-1913. These data are intended as a record of the 

 condition of the plaice population on the eastern side of the 

 Irish Sea during the years immediately preceding the war, and 

 they will enable us to make a comparison with the condition 



* Analytical methods can always be employed t(5 find the mode {dyjdx a 

 maximum on the summational curve), and the points of inflexion (dy-jdxi 

 maximal and changing sign on the summational curve). We think such 

 treatment would be pedantic. Measures of dispersion must be approximate 

 so long as we do not know the equation to the frequency curve. 



