288 Birge — The Crustacea of the Plankton. 



was so small as to make it inadvisable to insert it. In Figure 4 

 the change from bottle to thermophone is made in the last week 

 of July, and it will be seen that the lines come together with 

 great accuracy. 



Above the thermocline the bottle and thermophone agree 

 exactly, except at the surface on calm, sunny days, when 

 the reading of the thermometer is higher than that of the ther- 

 mophone, since by means of the thermometer the temperature 

 of a very thin stratum can be taken, while the thermophone 

 coil is of such a shape that it reads only the average tempera* 

 ture of a stratum some eight centimeters in thickness. 



During the period April — December, 1896, 189 sets of obser- 

 vations were made on 135 days varying from 3 to 6 per week. 

 In 1895, 196 sets of observations were made on 126 days in the 

 same period. 



The temperature observations were made at all hours of the 

 day; rarely by night, and must be taken as representing the 

 day temperatures of the water. Little difference, however, 

 would be made in the diagram if the night temperatures had 

 been introduced, as has been shown by an elaborate series of 

 observations made in 1897. Observations were regularly made 

 by single meters by the thermophone, and also by the bottle 

 when the difference between single meters exceeded one- half 

 degree C, and often when the differences were less. 



After recording the temperatures, those for meters not directly 

 observed were interpolated, and the average was taken of the 

 observations for each meter and each quar ter-month. 



In preparing Figs. 3 and 4 the average temperatures for each 

 meter and quarter-month were platted at the proper depth, and 

 in the center of the space representing the quarter-month on 

 the diagram. The position of the full degrees was then platted 

 on the assumption that a uniform decline of temperature is 

 found within a single meter. This assumption is incorrect in 

 the region of the thermocline as the zone of the most 

 rapid decline of temperature is frequently less than a meter in 

 thickness, but as this zone varies in thickness and shifts its 

 vertical position under the influence of the wind, little error 

 results from using this method of platting the average observa- 



