76 A Retrospect of Oceanography 



The density at temperature of observation can now at once 

 be reduced to its value at whatever is chosen as common tempera- 

 ture, and at the temperature which it had in situ. Naturally, 

 these reduced values are affected by whatever uncertainty 

 attaches to the tables used. For purposes of control, it is 

 well, in an expedition, to preserve very carefully considerable 

 samples of typical waters, and, as opportunity offers, to 

 determine their specific gravity against distilled water at 

 different temperatures. There is no reason to suppose that 

 the precision of these determinations would stand in any 

 way behind that of the observations on which the tables are 

 founded, and as they would have been made on the actual 

 waters which are under consideration, they should have the 

 preference. 



It will be obvious from these remarks how necessary it 

 is, in an expedition, to have a supply of perfectly pure distilled 

 water, to make parallel observations under the conditions 

 on board. Also, as before recommended, typical samples of 

 the waters collected should be kept for careful determination of 

 their density with the same hydrometer at different tempera- 

 tures, and especially at or near the temperature taken as the 

 common temperature of reduction. 



In the course of over twenty years' work with absolute- 

 weight hydrometers, and the determination of the constants 

 of about fifty different instruments, a curious fact has established 

 itself, namely, that the rate of change of volume of the instru- 

 ment with change of temperature is within certain limits 

 variable. The limits are very narrow, and the phenomenon 

 is to be detected only by very careful determination of the 

 displacement in distilled water. I at first thought it might 

 be due to distortion of the instrument with change of barometric 

 pressure; but many series of observations with different 

 instruments showed that the effect has no connection with 

 barometric pressure. For this reason, in the case of normal 

 determinations, parallel observations in the liquid under 

 determination and in distilled water of the same temperature 

 should always be made. 



The more use is made of an absolute-weight hydrometer, 



