138 ROYAL SOCIETY OF CANADA 



Montreal, who had paid particular attention to the insulation of the 

 cable, the fault was not located until late in the autumn (1907). Our 

 records are incomplete so far as the differential thermometers are con- 

 cerned from May to October (1907). Fortunately we succeeded in 

 obtaining thermograph records in the tower by placing a 10-day 

 thermograph of the usual pattern in the thermometer cage and mak- 

 ing weekly visits to the tower. In this way the temperature trace 

 at the tower could be compared with the daily thermograph records 

 taken at the Observatory, and the differences obtained. Calibrated 

 mercury thermometers were placed in the tower to set the thermo- 

 graph by each week, and to compare the trace with at each week end. 

 No record was made of the error of the thermograph during the week; 

 the mean between the beginning and end comparisons being taken 

 as the correction for the readings. In this way it will be seen that our 

 past summer records of the mountain differences are not so accurate 

 as tnose obtained from the differential recorder. We have, however, 

 decided to include them here since they are probably not in error 

 more than one or two degrees Fahrenheit. 



During October the mountain line was again made available by 

 renewing a small portion of the extra piece of cable, which had been put 

 in to bring the connections from the cable head up the tower to the 

 thermometer cages at the top. During all our trouble we least sus- 

 pected this wire, owing to its having been only recently installed. 

 It consisted of four No. 20 wires, rubber and braid insulation, encased 

 in a lead sheath. The wire had become very brittle, and an examina- 

 tion showed that it must have been "burnt " during annealing. After 

 removing the faulty pieces, records were resumed during part of 

 October, November and part of December. Further trouble was ex- 

 perienced the latter part of December, when it was found that the 

 low temperatures had further deteriorated the wire in the remaining 

 extra cable length. This decided us to remove all this faulty wire, 

 about 20 feet down to the cable head at the middle of the tower. 

 When this was done, our main cable was tested for insulation and 

 found absolutely perfect, even though the tests were carried out dur- 

 ing some mild and rainy weather in January. 



Owing to the difRculty of working on the top of the tower during 

 the severe winter weather, we decided to place the thermometer 

 temporarily in the lower box containing the cable head and connect 

 it directly with the wires from the cable. Our thermometer cage 

 being in a very flimsy condition we took advantage of the circumstance 

 to install a new cage which through the kindness of Prof. R. F. Stupart, 

 Director of the «Meteorological Service, was sent to us. At the same 

 time (March 30th) a new weather-proof four strand length, which 



