SEA-SURFACE TEMPERATURE. 
INSTRUMENTS AND METHODS 
Sea-Water Thermograph 
A continuous record of surface sea-water tempera- 
tures at a depth of approximately 2 meters below the 
surface was maintained by means of a mercury-in-steel 
bulb-and-capillary type sea-water thermograph with 24- 
hour movement (fig. 22). The recording apparatus for 
this instrument was located on a shelf in the chemical 
laboratory and communicated, through a lead capillary 
tube, with a large-volume mercury bulb in a protecting 
shield mounted on the hull of the vessel. The bulb was 
located 14.45 meters forward of the center of the rudder 
stock on the starboard side, 1.65 meters from the bot- 
tom of the keel by a vertical projection, and 1.71 meters 
out from the starboard edge of the keel. Under condi- 
tions of average draft, it was 2.29 meters below the sea 
surface. The thermometer bulb, itself, was 66 cm in 
length. Owing to the relatively small volume of mercury 
contained in the capillary tube as compared with the vol- 
ume of mercury in the bulb, considerable changes of 
temperature in the chemical laboratory produced no ap- 
parent effect on the recorded sea-water temperatures. 
The traces were changed daily, usually at noon (GMT). 
Canvas Bucket and 
Sea-Water Thermometer 
In order to control the thermograph records, the 
temperature of the surface sea water was measured by 
the bucket method immediately before each change of 
thermogram at noon. This method consisted of lower- 
ing a canvas bucket into the water until the upper rim 
was about 60 cm beneath the surface, then quickly haul- 
ing the bucket to the deck, and measuring, with a sea- 
water thermometer, the temperature of the water con- 
tained. The canvas bucket was approximately 30 cm in 
depth by 15 cm in diameter. The thermometer (P.T.R. 
No. 373) was a standard instrument and needed no cor- 
rections throughout the ranges of sea temperature en- 
countered on the cruise. 
Very few adjustments of the thermograph were nec- 
essary, as the difference between bucket and thermo- 
graph readings remained practically constant from day 
to day. Occasionally, however, there appeared to be a 
slow but steady change in this difference owing to un- 
known causes, and therefore it became necessary to re- 
set the recording pen of the thermograph on two or three 
occasions. In areas where the sea-surface tempera- 
tures were undergoing rapid changes, such as along the 
boundaries of well-developed ocean currents, or during 
calm, clear weather, these differences appeared to be 
somewhat erratic, owing probably to a lag in the record- 
ing mechanism of the thermograph. When the surface 
temperatures were changing rapidly, a mean of several 
bucket readings was used to determine the correction at 
that period. 
Figure 23 shows two interesting thermograms from 
the cruise. The upper trace (A) was obtained on the 
western edge of the Coastal Peru Current and indicates 
rapid variations of as much as 2°5 in about 10 minutes, 
presumably owing to the mixing of cold and warm water 
’ masses; the lower trace (B) shows the characteristic 
rapid changes of smaller amplitude recorded during 
calm, clear weather in the tropics. 
Evaluation of Thermograms 
The thermograms were scaled at each full hour, 
local mean time. The differences between the thermo- 
graph and bucket readings at noon (GMT) were deter- 
mined as has been described, and these values were 
used as corrections to the hourly thermograph readings. 
It is realized that the probability of error in values ob- 
tained by the bucket method is greater than in the case 
of individual values obtained from the thermograms [30] 
therefore the corrections to be applied have been 
smoothed considerably, except in instances where it is 
evident that the differences were due to a shift in posi- 
tion of the thermogram on the drum or to rapid changes 
in sea-surface temperature. At the lowest sea tempera- 
tures, the bucket thermometer readings averaged from 
0°8 to 0°9 higher than the thermograph temperatures, 
and at the maximum sea temperatures they averaged 
from 0°1 to 0°2 higher. Comparing sea-surface tem- 
peratures so obtained with those measured at each oce- 
anographic station (fig. 1) with the reversing thermom- 
eters, it is found that no difference greater than 0°5 oc- 
curred and at more than half of the stations this differ- 
ence was less than 0°1. 
DISCUSSION 
General Remarks 
When it is considered that the heat capacity of sea 
water is 3300 times greater than that of dry air at stand- 
ard pressureand temperature, it can readily be seen that 
the temperature of the sea surface controls, to a great ex- 
tent,-the temperature and vapor content of the overlying 
air. A knowledge of temperature conditions at the sur- 
face of the sea is therefore of fundamental importance 
to any study of marine meteorology. 
For these reasons, the observation and recording of 
sea-surface temperature was made a part of the routine 
meteorological work on board the Carnegie, and, as a 
result, corrected hourly values of sea-surface temper- 
ature are available for 330 days during the cruise. All 
days when the vessel was in harbor have been omitted. 
Mean Sea-Surface Temperatures 
for Groups 
The hourly values of sea-surface temperature given 
in table 79 of appendix III have been summarized for the 
Groups outlined in table 1 and figure 3. These Groups 
were originally defined by Miss Clarke, and were con- 
structed mainly on the basis of homogeneity of sea-sur- 
face temperature. Unfortunately, it is impossible to 
designate arbitrary geographical boundaries on the sur- 
face of a constantly moving and changing sea, and to ex- 
pect the areas described by these boundaries to be true 
climatological entities throughout any given period. It 
has been necessary, however, to divide the data region- 
ally in some manner for purposes of analysis, and it is 
believed that Miss Clarke’s classification should serve 
this end. The hourly values corrected for noncyclic 
change, and the mean sea-surface temperatures for each 
27 
