14 Universitij of California Publications in Zoology [Vol. 15 



01- three minutas later the thermometer is read and the temperature 

 estimated to the nearest 0?05 C. Owing to the fact that the scale of 

 these thermometers is made of paper so graduated that 0.2 of a degree 

 is represented by a di.stance of only 0.71 mm., it is evident that esti- 

 mates closer than 0.05 would be unreliable. After reading the tempera- 

 ture, a pint of water is tightly sealed in a '.'sure-seal" fruit jar, 

 properly labelled, and filed in the naturalist's house until it can be 

 taken to the laboratory and analysed. In making the final record of the 

 temperatures the readings are corrected from a table of calibration 

 errors previously obtained by comparing the thermometer with a 

 standard from the "Reichsanstalt. " 



(&) Sub-snrface samples. — Prior to July 16. 1910. all the sub- 

 surface samples were collected with the Kofoid water bottle (see p. 

 11). The temperatures were taken with the same thermometers and 

 the samples were labelled, preserved, and filed as in the case of the 

 surface samples. However, owing to the fact that the Kofoid bottle 

 is made of bronze and is not insulated, the friction of the surrounding 

 water on the bottle as it is being hauled up, in addition to the conduc- 

 tion while in the upper layers of water, appreciably increases the tem- 

 perature of the enclosed sample. In fact, observations show that the 

 water in the top may be from two to three degrees warmer than that 

 in the bottom of the bottle, when the samples are obtained from 

 depths exceeding three hundred meters. For this reason the tempera- 

 ture of the water in the bottom of this bottle was always taken. 



During July, 1912, several series of parallel temperature mea.sure- 

 ments were made, one set by the above method, and the other with the 

 Ekman reversing water bottles provided with Richter reversing ther- 

 mometers (see p. 15). Prom a comparison of the results the effect 

 of the friction and conduction above mentioned was ascertained and a 

 correction for the same relative to depths less than 350 meters was 

 computed. No reliable correction could be made for greater depths, 

 owing to the fact that the increase in temperature due to the extra 

 friction and conduction is almost, if not quite, balanced by the decrease 

 in tempei-ature due to the increase in depth. The following list of 

 corrections shows the amount by which the temperatures of the Kofoid 

 bottle miLst be decreased in order to obtain the actual temperatures at 

 the depths of closure : 



