CLYDE SEA AREA. 



673 



discrepancies in the two density determinations are greatest the depression produced by 

 a given weight is least. Thus it is probably some cause which prevents the weight from 

 exerting its full power, or which enables it to act more strongly than it would unaided, 

 that produces unusually high or unusually low discrepancies. Errors in readings would 

 also partially account for the range. The cause was not discovered, but many observa- 

 tions pointed to a difference in capillary attraction due to differences in the cleanness, 

 and consequently the surface tension of the liquid. Most of the exceptionally high 

 depressions occur, as might be expected, in the case of exceptionally fresh samples of 

 water, and before using the indications to measure the apparent error of density deter- 

 minations these should be excluded. Table XVII. gives the result of the 598 determina- 

 tions which were investigated without excluding any of the cases. 



Table XVII. — Summary of Hydrometer Variations. 



Deviation from Mean Depression. 







0-5 



TO 1-5 



2-0 



Times in 598, 

 Times per cent., 



195 

 32-7 



232 

 39-0 



105 

 17-5 



48 

 8-0 



17 



2-8 



Thus in 71'7 per cent, of the cases the reading differed from the mean value by half 

 a millimetre or less, which corresponds to 0'00003 in the density of a sample of sea 

 water of ordinary salinity. The less favourably conducted work on the Clyde Sea Area 

 was liable to a somewhat higher uncertainty, but this scarcely exceeded 0*00005, so that 

 the densities may be relied upon to the fourth decimal place, and for purposes of com- 

 parison they are not meaningless in the fifth. 



In order still farther to test the accuracy of the hydrometer, and to provide a 

 standard by which the salinity deduced from its readings may be compared with 

 Dittmar's work, the chlorine determinations of Mr Dickie* have been employed. The 

 value of x> *- e - total halogen considered as chlorine, in grams per litre, in 96 cases was 

 written down, and the value ( 4 S£) of the density of the water deduced from Dittmar's 

 table was set opposite it. The actual density determined by hydrometer on the "Medusa" 

 in a duplicate sample of the water was then taken, and the difference between this and 

 Dittmar's theoretical density noted. Out of the 96 cases the densities were identical 

 in 3, the observed density was higher than the calculated in 59 cases, and lower in 

 34. The average deviation, irrespective of sign, was '000 11 3, or considering the 

 sign the average deviation was an excess of the hydrometer — determined over the 

 calculated density of 0*000024. The observed densities in these cases were determined 

 on board the " Medusa," and being thoroughly typical of the whole Clyde work show, 



* Proc. Roy. Soc. Edin., vol. xiv. p. 422 ; vol. xv. p. 283. 

 VOL. XXXVI. PART III. (NO. 23). 



5 I 



