19G 



Appendices to Fourth Annual Report 



Tube I. . . Mean of 



„ II. . 



„ III. . 



„ IV. . 



„ V. . 



„ VII. . 



„ VIII. . 



. „ X. . 



„ XIL . 



Tubes I. and IV. are evi 





Difference 





in mgrms. 



4 



n.OQ 



3 



Oil 



5 



A.I A 



4 



0-38 



3 



0-10 



5 



0-29 



6 



0-18 



4 



0-15 



5 



0-19 



ideiitly singled out as being in some way defec- 

 tive. The only points noticeable in those two tubes were, first, that owing 

 to a very slight irregularity at the end of the capillary terminal, the flow of 

 the water on applying the blotting paper was impeded, which made the 

 rapid adujstment of the level of the water in the wider terminal some- 

 what difl&cult, and thus increased the time during which it was necessary 

 to keep the tubes out of the melting ice ; and, secondly, that in these 

 tubes the bore of the wider terminal was greater, than in the other tubes, 

 and, cG&teru paribus, the effect of the observational error in adjusting the 

 level of the water in the wider terminals of these tubes must have been 

 greater than in the other tubes with wider terminals of narrower bore. 

 These two very slight defects fully account, I believe, for the inferior 

 results obtained with these two tubes, though they were not sufficient to 

 condemn them at the first trials, and before all the results could be com- 

 pared with each other. 



Passing from the consideration of the weighings as such to the specific 

 gravities derived from them, it will be observed that only in two cases out 

 of thirty does the difference between the two corrected specific gravities 

 reach 2 in the fifth place of decimals, viz., in Tube VII., station 22, 20 

 fathoms from surface, where the difference in the fifth place is 2 '2, and in 

 Tube VIII., station 54 bott., where the difference is 2'0. Even on the 

 lowest assumption, i.e., that the difference in a given case is to be credited 

 to one of the determinations, and not to be regarded as arising from the one 

 result being too high and the other too low ; the specific gravity must still 

 be regarded as correct to about 1 in the fifth place of decimals. This is 

 most satisfactory. None of the methods hitherto adopted for hydro- 

 graphic work can claim an equal accuracy. Moreover, in thus taking an 

 ex.treme case, we are evidently undervaluing the accuracy of the method. 



The only two instances in which the differences reach 2 in the fifth 

 place were not obtained, as might have been expected, with the defective 

 Tubes I. and IV., but with Tubes VII. and VIII. In these two cases the 

 differences between the corrected weights are 0*53 mgr. and 0*44 mgr. 

 respectively, — differences which, even if they be not classed as accidental 

 certainly have a disproportionate influence on the specific gravities, owing 

 to the comparatively small capacity of these particular tubes. If in future 

 work only such tubes as have a capacity of circa 30 to 35 c.c. be em- 

 ployed, the effect even of such differences would be very much reduced. 



In order to facilitate reference to the annexed chart, I have arranged in 

 Table III. the ' corrected average specific gravities ' in order of magnitude 

 under the heading qS^.* The numbers in this column are for the most part 

 the same as those given in the last column but one of Table II. They 

 are strictly comparable with each other without any correction, and are 

 derived directly from my observations. In order to facilitate comparison 

 * ^S^' means specific gravity at t"^ referred to water at f as = 1000. 



