32 rnivcrsifij <i{ Califdrnia Publications in Zooldijij [Vol. Id 



"While the mean error, as shown hy this list, is only about O.OOOOo, 

 that actually met with in practice was fully twice as large, and oc- 

 casional eiTors as large as 0.00025 were found. 



We had hoped, by employing these hydrometers and occasionally 

 checking the results with a pycnometer, to reduce materially the time 

 and still maintain the requisite degree of accuracy. While it has been 

 possible to make twelve observations per hour, the saving in time is 

 more than offset by the loss in accuracy, so that the method was 

 discontinued. 



(c) Mohr's litrimrtric mvtliod. — This method wa.s first used in 

 the summer of 1910. It is a volumetric method by which the chlorine 

 or, more correctly, the halogen content in a definite volume of sea- 

 water is precipitated by a measured volume of a solution of silver 

 nitrate (AgNOg) of known strength. The strength of this .solution 

 may be determined in several ways, but the one commonly used and 

 upon which we relied entirely is that of titrating a sample of "normal 

 sea-water" of known chlorine (halogen) content. This normal water, 

 as well as the apparatus used, was obtained from the Central Labora- 

 tory of the International Commission at Christiania. In titrating, 

 yellow potassic chromate (K._,CrO.,) is used as an indicator which, 

 reacting on the silver nitrate after all the halogens are precipitated, 

 produces the red precipitate of silver chromate (AgoCrO^). For 

 further details concerning the method and apparatus used see Helland- 

 Hansen, 1911-12, pp. 34-39. 



After several weeks' practice, a series of water samples taken from 

 the ocean and from San Diego Bay were carefully tested by this 

 method and compared with measurements of the same samples made 



with the pycnometer. The results, reduced to specific gravity iS- 



•4 



by means of Knudseirs tables, are listed on the opposite page. 



While the mean error, as showTi by this list, is about 0.00007. it 

 has been occasionally found as large as 0.00020. Furthermore, the 

 magnitude of the error is so erratic that the method has. thus far, 

 proven unreliable. In addition, the time required has been fully as 

 great as with the pycnometer. and the expense involved is much 

 greater. The method was, therefore, discontinued. 



{d) The sinker method. — Eepeated attempts were made to find 

 a method of determining salinity which would combine the accuracy 

 of the pycnometer with the speed of the hydrometer. These efforts 

 culminated in 1912 when a method depending on the apparent w-eight 



