REPORT 1902-03 APPEND. A — XLII — 



determination , which cannot be controlled. The old principle of insulating is 

 therefore adhered to; however, the avoidance of direct metallic connections 

 between the outer and inner parts has necessitated several alterations in the con- 

 struction of the new water-bottles. 



The present model insulates to within 0.01° for 8 or 9 minutes, if the 

 difference of temperature outside and inside be 10° G. But after this the tempe- 

 rature begins to change very quickly. The water-bottle gives the correct temperature 

 (to within 0.01°) if notice be taken that the reading does not change more than 

 0.01° per minute after the reading is taken, and if the reading be not taken more 

 than 10 minutes after the water-bottle is closed. 



The large Pettersson-Nansen water-bottle is not made any more, because 

 the thermal effect of pressure upon its walls, when the bottle is used at great 

 depths at which the smaller one cannot be used, gives rise to appreciable errors. 

 On account of these thermal effects it is necessary at great depths to determine 

 temperature by means of reversing thermometers. 



Such instruments have latterly been very well made by Richter, Berlin. 

 The C. L. has had several of these thermometers and they have proved able to 

 allow measurements of the highest accuracy required. In melting ice the mercury 

 breaks off accurately at the same point, to within smaller differences than could 

 possibly be observed by careful reading under ordinary circumstances. (The 

 different readings do not differ by more than 1/20 or V40 of an interval on the 

 scale). The auxiliary thermometer*) which is enclosed in the outer tube of the 

 instrument and gives the temperature of the thread of mercury broken off', allows 

 the corresponding correction to be determined with an accuracy of 0,01° to 0,005° 

 — if two or three readings be taken — without making use of a water-bath. 

 According to experiments made in the G. L. as well as on the sea, the thermo- 

 meters seem to be free from the defect of letting down small drops of mercury 

 after the reversing of the thermometer. These reversing thermometers therefore 

 allow of the same accuracy in the determination of temperature as does an ordi- 

 nary thermometer of similar graduation. Further, the readings are free from any 

 error arising from the compressibility of solid bodies used in the construction of 

 the instruments, or of the sea-water itself. In consequence the temperatures 

 found by reversing thermometers and by insulating water bottles, cannot be com- 

 pared with one another directly. It is consequently to be noted that all tempera- 

 tures determined by insulating water-bottles, should be corrected for the cooling 

 caused by the decrease of pressure. 



*) It may be mentioned here, that reversing thermometers witliout such auxiliary thermo- 

 meters do not allow any accurate determinations, unless they be read off in a water-bath 

 or some similar jacket of kno^vn temperature. 



