40 



University of California Publications in Zoology [Vol. 15 



Attention is called to the fact that these differences are in every 

 ease in the same direction, B's determinations always being higher than 

 A's. It is therefore evident that the titration method is subject to a 

 systematic error due to the "personal equation" of the observers which 

 is nearly twice as great as that admitted to be inherent in the method. 

 This is probably due to the difference in judgment of color for, as 

 stated by Bjerrum (1904. p. 8) : "The greatest annoyance in the 

 titration with chromate of potash as the indicator, especially to unex- 

 ercised people, is that even a strong tint may disappear on fast 

 stirring for some time." 



It would seem, from the above discussion, that ]Mohr's titrimetric 

 method is, as a rule, eaiiable of a high degree of accuracy. However, 

 since in such cases the determinations have been made on samples taken 

 from a region strictly comparable to that from which the "normal 

 sea-water" is obtained, our own failure to attain a degree of accuracy 

 comparable to the pycnometer method may in part be due to a devia- 

 tion from a normal proportion of the salts in the water about San 

 Diego (see p. 41). Nevertheless the titration method is gaining 

 ground in oceanography, not because it can replace the pycnometer 

 in accuracy, because the latter is used for standardizing all other 

 methods, but for the reason given by Helland-Han.sen (1911-12, p. 

 39), namely: "it is so rapid that it makes possible the consecutive 

 treatment of a great collection of water samples, and so accurate that 

 none of the other tolerably fast methods bear comparison with it." 



