DISTRIBUTION OF SURFACE TEMPERATURE iQS 



TREATMENT OF THE DATA 

 As mentioned above, the continuous thermograph provides the most important body of data. In this 

 instrument the thermometer bulb Hes in a pocket in the ship's hull about 14 ft. below the surface, 

 and the temperature at that depth is recorded on a chart (see Fig. 5, p. 187) attached to a clockwork 

 drum in the laboratory on deck. Each chart covers a week. The instrument has given very satisfactory 

 results and, provided certain checks are made, the temperature can be read correctly to within less 

 than 0-2" C, and the time to within less than half an hour. Certain possible sources of error have to 

 be considered. The clock has generally been found to keep excellent time, but allowance had occasionally 

 to be made if the drum had not been accurately set when the new chart was attached. The timing could 

 generally be checked at points where the ship was stopped on station, for here the temperature is often 

 traced as a steady horizontal line which is usually distinguishable from the numerous small fluctua- 

 tions of temperature which appear while the ship is moving (for example, see Fig. 5 A, p. 187, St. 451). 

 The time when the ship is stopped can of course be checked from the logs. Correction of the tempera- 

 ture is more important as there is frequently a slight error which is possibly due to a small amount ot 

 play on the arm of the pen. This error seldom exceeds 0-5^ C and is usually constant until the chart is 

 replaced unless substantial changes of temperature occur. The thermograph record should, however, 

 always be checked against direct thermometer readings, preferably with the Nansen-Pettersson water- 

 bottle Since the ' Discovery ' and ' Discovery II ' normally worked full stations at frequent intervals, 

 there were few thermograph charts which could not be checked in this way at several points, but it is 

 verv evident that the method of reading the temperature in a bucketful of water drawn over the ship s 

 side is wholly unreliable as a check, however carefully the thermometer is read. Such readings seem 

 nearly always to be too high and the error may amount to i-o° C or more. This method has been 



criticized elsewhere (see Lumby, 1928). , , r r. 1 



A further question which arises is whether the thermograph, recordmg at a depth of 14 ft-, Proper y 

 shows the 'surface' temperature, and whether thermometer readings at o m. constitute a valid check. 

 In other words, what diflFerence may there be in Antarctic waters between the temperature at o m 

 and about s m.? At full stations the vertical series of temperatures, etc., included readings at o and 

 10 m and I have worked through the station lists and assembled the data on this point m Table 8. 

 This includes 987 vertical stations at which the temperature at o m. did not exceed 5-0 C (i.e. the 

 vast majority of stations south of the Antarctic convergence) and it will be seen that, except in the 

 summer months (December-February) about 90% showed less than o-z^ C difference between o and 

 10 m Even in the summer months when occasional superficial sun-warmed patches may be expected, 

 any significant difference is rare, and the difference between o and 5 m. must be still less. There have 

 beL one or two occasions when a vertical station has been worked in the vicinity of an iceberg, and 

 where a cold, less saline, surface stratum has depressed the temperature at o m. below that at 10 m 

 In any case where there is a significant diflference between o and 10 m. it is preferable to check the 



thermograph chart from other stations. , , 1 . f*^„ 



Since ship's time changes with changing longitude the clocks are frequently changed at sea, often 

 several times in a week. The thermograph cannot conveniently be adjusted to ship s time and is there- 

 fore set at G.M.T. But the temperature readings must be linked with the ship s position the stations, 

 nd her observations, all of which are recorded by ship's time. It is therefore desirable to mark a 

 "ak of ship's time on the thermograph charts. I found it best to trace the corrected temperature on 

 to new charts, marked with the ship's time scale, periods on station, and other -notations. 



The next step was to draw track charts, marked with ship's time, for all voyages south of 45 S 

 except such complex local movements as were undertaken in the oceanographic surveys around South 



