Temperatures. 
287 
line, and allowed to fill. It was then drawn rapidly to the sur¬ 
face and the temperature read by means^of a long-stemmed ther¬ 
mometer graduated to one-fifth of a degree. The time of rais¬ 
ing the bottle from the bottom of the lake was ordinarily about 
ten seconds; and the small size of the opening prevented mix¬ 
ture of the upper water with that in the bottle. The tempera¬ 
ture of the water in the center of the bottle, which was meas¬ 
ured by the thermometer, did not change perceptibly during the 
time required for the thermometer to set. The water from the 
lower part of the lake, however, was somewhat warmed by con¬ 
tact with the glass and the air in the bottle. This error was 
carefully determined by comparison with the thermophone, and 
is about one-fifth of a degree C., when the difference between 
surface and bottom is about 10 degrees. 
Errors much more considerable than this occur with the use 
of the temperature bottle at the thermocline. In this region 
the temperature may fall as many as nine degrees in a single 
meter, and not infrequently as much as three or four degrees 
in a quarter of a meter. It is impossible that the bottle should 
take in all of its water from the stratum in which its mouth lies 
as the escaping air sets up currents so that a mixture of the 
water occurs. A difference of half a degree may therefore oc¬ 
cur between the readings of the thermophone and the bottle in 
this region. In one case the error amounted to two degrees, 
where the bottle was opened a few inches below the upper level 
of the cold water and took in a mixture of this water with the 
lower part of the warm stratum above. The errors at this re¬ 
gion, however, while considerable, make little difference in the 
average results of observations, since their only effect is to 
make the upper level of the cold water appear to be a fraction 
of a meter lower than it really is. Since this level is subject 
to irregular variations, under the influence of the wind, which 
may amount to two or even more meters, the errors introduced 
by the bottle are insignificant in the average of a week’s read¬ 
ings. It was intended to correct the observations of the bot¬ 
tle by means of the thermophone and to introduce the correc¬ 
tion in the diagrams of temperature. It was found, however, 
that the amount of correction to be introduced in the diagrams 
