332 



Part III. — Tiventieth Annual Report 



furnished with mats and an oil-stove, it is fairly comfortable, and has 

 been found of great service in cold and stormy weather. It is lit at night 

 either with acetylene gas led from the ship's apparatus, or by a lamp. It 

 is strong enough to withstand the wind and a good lashing of spray ; when 

 seas are shipped the canvas is unrolled, but the pitching of the vessel then 

 makes work impossible. 



Fig. 3, — Portable Deck-house. 



In many cases when it was found impracticable on board to measure 

 the fish from the small-meshed net, they were jilaced in air-tight tanks in 

 a weak solution of formaline and transferred to the laboratory, where the 

 measurements were made, These tanks, one of which is shown on the 

 left of the cut (fig. 3), have been found very convenient for this purpose 

 and for the preservation of other material. 



In dealing with the results graphic diagrams or curves are made of each 

 series of measurements, the grouping varying according to the sizes of the 

 fishes. For most round fishes, the larger flat-fishes, skates, and rays, one- 

 centimetre groups usually suffice ; but even in such cases it is often 

 necessary, particularly with the older series, to plot them out also in "S-cm., 

 or even 2- or 3-mm. groups in order to determine the limits of the series. 

 In some cases, e.g., gurnard, where the coalescence of successive genera- 

 tions is marked, one must often group the mm. measurements in a variety 

 of ways before feeling satisfied as to the point of division. In the case of 

 the smaller flat-fishes, and smaller fishes generally, -b-cm. groups are much 

 more satisfactory for the curves. 



Having fixed upon the point of division — in the earlier series usually 

 quite a simple matter — the average size of a series was obtained by the 

 computation of the measurements comprised within it, which gave the 

 arithmetical mean. This method, besides being laborious, does not always 

 give the true mean size of the series unless the fishes of different sizes 

 are equally represented, and a simpler and better method is to take the 

 median point between the extreme limits of the series, when these limits 

 have been well ascertained. Sometimes considerable difficulty was 



