1915] Michael, rt al. : 11 [/droyrapltic Records of Scripps I tisl il iil inn 25 



closing-net hauls made with the same net from 50 to 40. 40 to :^0, 

 30 to 20, 20 to 10, and 10 meters to the surface. 



(c) Preservation of water samples. — Troublesome errors in salinity 

 often result from the manner in which water samples are preserved. 

 In fact, if the samples are not tested soon after being collected, ab- 

 normally high salinities will occasionally occur despite the utmost 

 care in sealing the containers. While the amount of evaporation has, 

 in some cases been great enough to be detected veadily, it has generally 

 been so slight as to require a special examination of all our salinities 

 for its detection. 



f° 

 Such an examination has shown that the density in situ, S—^, rela- 



4 



tive to water samples collected at approximately the same time from 

 a series of depths, nearly always increases but never decreases as the 

 depth increases. This accords with the experience of Makaroff (1904. 

 p. K:!!)! who says: "Les poids specifiques doivent augmenter avee la 



t° 

 profondeur; s'il arrivait que le poids specifiqile >S' -^ de I'eau de 



4 



profondeur etait inferieur a celui qui serait observe dans les couches 



superieurs, on pourrait en deduire 1 'existence d'une erreur dans la 



determination du poids specifique. " 



This empirical relation of densities in situ, which Makaroff and 

 others, as well as ourselves, have found by observation, has also a 

 theoretical basis. Stability of the water layers coiild not otherwise 

 exist. 



The actual density of a mass of water at any depth depends upon 

 its temperature and pressure, as well as upon its salinity, the relation, 

 of course, being such that the density decreases with a reduction in 

 pressure and increases with a reduction in temperature. Now, when, 

 ov^^ng to instability, a mass of water at any level a rises to a higher 

 level b, its densit.y changes. The expansion of the water mass due to 

 the reduction in pressure directly decrea.ses its density but, as it also 

 slightly decreases its temperature, the actual decrease in density is 

 not quite so great as it otherwise would be. However, since the reduc- 

 tion in temperature due to an ascent of even five himdred meters is 

 less than 0?05 C. and the corresponding increase in density less than 

 0.00002, while the decrease in density directly caused by the expansion 

 is 150 times greater, only the latter effect need be considered. Now, 

 while the density of the rising mass of water is being coutinuou.sly re- 

 duced owing to the decrease in pressure, both diffusion and conduction 



