34 DYNAMIC METEOROLOGY AND HYDROGRAPHY. 



2 C, to the maximum values in the tropics, hardly anywhere exceeding 30 C. 

 in the open sea. As we proceed downwards, the freezing-point of sea-water is 

 retained as the lower limit, while the upper gradually decreases, but at a very differ- 

 ent rate, in the open ocean and in the more or less closed seas. The temperature 

 in the open sea will hardly anywhere exceed io C. at a depth of 1000 meters, and 

 in still greater depths it will be found between the limits +2 and 2. In more 

 or less closed seas the temperature may be much higher in corresponding depths. 

 Thus, in the Red Sea there is a temperature of 2i.5C. at the depth of 2100 

 meters, in the Sulu Sea (between Borneo and Philippines) 10.2 C. at the depth 

 of 4300 meters, and in the Mediterranean 13.9 C. at the depth of 4400 meters. 

 The salinities also have their greatest range of variation at the surface, namely, from 

 zero at the mouths of great rivers to 39/ o hi the Mediterranean and 40.4 /oo m the 

 Red Sea. As we proceed downwards the low salinities rapidly disappear, but at 

 a different rate and converging towards different limits in the open ocean and in 

 the closed seas. In the open ocean the salinity rapidly converges towards the 

 almost constant salinity of about 35 / 00 . The Mediterranean has the higher salinity 

 of about j 39/oo and the Red Sea of about 4o/ 00 at the bottom, while the Baltic has 

 the low salinity of about 10 / 00 in its greatest depths, somewhat exceeding 400 

 meters. Lower salinity than that of the Baltic and higher than that of the Medi- 

 terranean or the Red Sea will hardly be found anywhere in corresponding depths. 

 These general data have determined the limits of the three last tables. 



The method of tabulation which we have used, while leading to very small and 

 convenient tables, has a defect which must be mentioned. The result is found as 

 the sum of seven terms. Each term may have an error of five units in the sixth 

 decimal of the specific volume, or, on account of the double interpolations in the 

 tables containing more than one variable, even somewhat more. Thus in excep- 

 tional cases errors may occur exceeding 3.5 in the fifth decimal, corresponding to 

 an error of about 0.03 /<x> m the salinity and thus exceeding somewhat the errors 

 of careful observations, which may be carried to about 0.01 / 00 . This error may 

 perhaps be of importance for the investigation of the conditions of equilibrium or 

 motion in very homogeneous parts of the sea. It might have been avoided if we 

 had calculated all our tables with one decimal more. But this would have made the 

 use of the tables in the most common cases much less convenient. When more 

 accurate tables are required it will probably be the best plan to construct three 

 different sets of tables for three different types of sea-water: the oceanic type, the 

 Mediterranean type, and the Baltic type. For the oceanic type the tables should 

 be extended to all pressures, but for all greater pressures the temperature and 

 salinity could be contained between very narrow limits, approaching C. for tem- 

 perature and 35 /oo for salinity. For the Mediterranean type the tables should 

 extend to pressures somewhat exceeding 4000 d-bars, while the temperature in 

 greater depths has about 10 for its lower and 20 for its higher limit. The salinity 

 in greater depths should have 35 for its lower and a little more than 40 for its higher 

 limit. These tables might be used in closed tropic or subtropic seas, as the Medi- 

 terranean, the Red Sea, and the Sulu Sea. For the Baltic type the tables could be 



