142 ROYAL SOCIETY OF CANADA 



it was thought desirable to devote a considerable amount of time to a 

 determination of the zero of the instrument under varjdng conditions of 

 weather. For this purpose it was necessary to operate the recorder with- 

 out the thermometer bulbs in their respective places. The long lead8 

 running to the top of the mountain, a distance of over 4100 feet were 

 therefore connected together as well as the compensating leads at the top 

 of the tower. At the observatory the thermometer bulbs were joined in 

 series ^^'ith these wires, where they were connected to the re- 

 corder, and were placed side by side in a vessel of water in order to keep 

 them at equal temperatures. It was not found necessary to immerse the 

 thermometers in ice to obtain the zero since the zero was merely the point 

 of equal temperature at the bulbs. On account of the care taken to 

 adjust the two thermometers to equality at the outset no appreciable 

 shift in zero took place when the point of equal temperature was shifted 

 along the temperature scale. 



A test of this kind ranging over a long period is of the greatest im- 

 portance in determining how adequate the compensating leads are in cor- 

 recting for changes in temperature in the main leads. We had pre- 

 viously noticed when determining the zero in this way that small varia- 

 tions took place, resulting in a slightly zigzag line in place of the perfect- 

 ly straight line, which is obtained when the recorder is connected to ratio 

 coils of equal resistance. In the main it is possible to distinguish these 

 small irregularities from the waves produced in the trace by small 

 changes in temperature. 



In the present zero tests we noticed decided fluctuations which ap- 

 peared both with the thermometer bulbs connected, and when we re- 

 moved them entirely from the circuit. These irregularities were as 

 before of the order of one degree on either side d"f a central line which 

 we took as the zero. Sometimes the pen would remain to the 

 right of the line for hours tracing its characteristic line, and at. other 

 times we obtained the trace about the same amount to the left of the 

 zero. In many cases we observed that the fluctations Avere more rapid 

 and lof greater extent, and the pen moved back and forth in half hourly, 

 or hourly periods. 



In general we found that during the night we obtained the steadiest 

 zero traces, and in particular when there was no Avind. During a day 

 with bright sunshine and high wind the trace was mos't unsteady. The 

 approach of sunrise was in nearly every case accompanied by a slow 

 displacement of the pen to the right of the zero. The variations noticed 

 with a high wind would be accounted for by strains set up in the cable 

 which would act in an irregular manner on the various wires, and the 



