55 



The measurements are therefore not exactly the same as those 

 of Heincke and other previous investigators, but it is easier to 

 measure by means of a mechanical board along parallel lines, and 

 it is more than likely that differences due to degree of development 

 are to a large extent obviated by this method. 



In the time at our disposal it was found impossible to give the 

 figures the analysis which they certainly deserve, but as an effort 

 is being made at present to determine the races of herring the whole 

 of the tables have been sent to the Board of Agriculture and Fisheries 

 in order that they may be treated in parallel with those obtained 

 from other sources. 



What has been done is this — and it will be found to be an 

 adaptation of a scheme for rapid analysis given in the report for 

 1903, under the heading " A Contribution to our Knowledge of 

 the Growth of Plaice " — to bring together the whole of the measure- 

 ments on a chart. The method has the advantage of determining 

 rapidly whether any difference is exhibited, the variations, obvious 

 mistakes, and especially in the case of differences, the measurements 

 which deserve more careful mathematical investigation. 



Briefly the method is as follows : A basal measurement, pre- 

 ferably total length is taken as the basal or " x " Hne, and all other 

 measurements of the sample or groups of samples are expressed on 

 the chart by dots with reference to the basal measurement. In 

 the case of the herring measurements, as it was believed that the 

 total length was not to be relied upon with the same confidence as 

 the length of the scale-covered portion of the body, the latter 

 measurement was chosen as the one in terms of which the others 

 were to be expressed. 



If this be done it will be found that the measurements practi- 

 cally resolve themselves into groups along Unes originating in o. 

 After the sample has been thus charted, a line is drawai as nearly 

 as judgment will permit through the groups from o, and it is easy 

 then to express the results as tangents of the angles thus obtained. 

 In the table which will be submitted presently the tangents were 

 calculated by noting where the respective lines cut the upright 

 from the 22 cm. measurement of the basal line. 



