The of the air was determined colorimetrically (McClendon, 1917), 30 observations showing it to vary from 2.8 to 3.5 parts per 10,000. A new aneroid barometer was used and the mean value for 2 months was taken as 760. A thermometer graduated down to tenths of a degree was standardized by the United States Bureau of Standards and the other thermometers compared with it. A recording ther mometer for the air was compared with the mercury thermometer, but the records were not found to be of much value in relation to the sea. Tortugas is about on the heat equator during July. There was a diurnal variation of the temperature of the air from 25° at dawn to 30° at 1 p. m., with a very slight upward drift during July and with a few variations due to storms. All time records are in local apparent time. The weights used were standardized by the United States Bureau of Standards and the glass apparatus and solutions stand ardized by weighing.
All water samples that could not be examined immediately were preserved in large flasks of resistance glass entirely full of the sample and were examined as soon as possible. Most of the samples taken at night were titrated by artificial light. The pH was in such cases deter mined with the aid of a "daylite" lamp, but the results were not very satisfactory.
Samples (for pH) taken a short time before sunrise were kept in nonsol flasks for the pH to be determined after sunrise.
tension is one ten-thousandth of a normal atmosphere (760 mm. Hg) of The sea-water was usually supersaturated or undersaturated with oxygen. The number of cubic centimeters of 02 per liter at saturation at a given temperature was read from figure 5 and the excess or defi ciency of saturation of the water was recorded. Figure 5 com pares favorably with the results of Fox, who used a different method of analysis. Jacobsen's formula for cubic centimeters 02 per liter at saturation is 10.062 — 0.2822 t —0.006144 — 0.000061 —0.10'73 Cl +0.003586 tCl — 0.000055 t= temperature and Cl = grams chlorine per kilogram of sea-water.
The determinations for constructing this curve were made as follows: A wash-bottle of about 500 c.c. capacity was supplied with about 400 c.c. of sea-water and immersed in a thermostat. Moistened out door air was drawn through the wash-bottle by means of a suction pump until equilibrium was established. The air passed through a very small wash-bottle of sea-water and a copper coil (both immersed in the thermostat) before reaching the 500 c.c. wash-bottle. The analyses were made by the Winkler method, using the same standards as with the other determinations, a correction being applied for the average gas-pressure in the wash-bottle. No correction was made for the vapor tension of water, since it was the same as at the sea surface everywhere.