30 Universiiij of California Publications in Zoology [Vol. 15 



where 6 is the temperature of the water sample at the time of measure- 

 ment, and w the apparent weight, in grams, of the pycnometer filled 

 with the water sample. If the temperature, 61, lies between 10° and 

 30° C and is read to within 0?03, and if the weight iv is determined 

 to within one milligram, it is possible to obtain the specific gravity, 



e 



S -r to an accuracy of 1 in 100,000. 

 4 



"While this accuracy is po.ssible. it is rarely obtained in practice. 

 Small air bubbles may at times escape notice, the temperature actually 

 read may not represent, with sufficient accuracy, that of the sample 

 at the time the pycnometer is closed, or several other soiirces of slight 

 error may be iinconsciously met with when a large number of samples 

 are being tasted which are avoided during calibration tests. 



In order to reduce such errors to a minimum, two pycnometers, 

 independently calibrated, were alternately used until about two hun- 

 dred samples had been tested, when the calibration constants were 

 checked by observations on distilled water. As an additional check 

 the same samples were occasionally tested by both instruments. Pro- 

 ceeding in this way, an average accuracy of 2 in 100,000 was obtained, 

 although in some instances the error was greater. The most serious 

 objection to this method is the time required which, including the 

 necessary computations, amounts to about ten minutes, in spite of the 

 fact that tables were compiled to facilitate the computations. 



(ft) The hydrometer method. — In the fall of 1908 three sets of 

 hydrometers made by R. Kiichler, of Ilmenau in Thiiringen, were pro- 

 cured. The five instruments in each set range respectively from 1.000 

 to 1.007. 1.006 to 1.013, 1.012 to 1.019, 1.018 to 1.025, and 1.024 to 

 1.031. Their volumes range from 110 to 130 e.c. and the volume of 

 their stems corresponding to one scale-division is 0.0001 of that of the 

 hydrometer. The stems are about four millimeters in diameter, the 

 length of one scale-division being about one millimeter. 



When using these instruments, the place on the stem cut by the 

 level water surface is read and care taken to have the instrument and 

 especially its stem clean. The hydrometer is allowed to oscillate and 

 is read just after coming to rest, the stem thus being wet for a slight 

 distance above the water surface. The temperatures are estimated 

 to 0?05 C by a floating thermometer graduated to 0?2. 



The readings of the hydrometer are, of course, corrected for the 

 variation in temperature at the time of observation and two separate 

 determinations are made of each sample, the results obtained being 



