In A of this illustration, the calibration tank was held at a constant tempera- 

 ture of 75.00°F ±0.005° while recording I was made, then lowered to 70.00°F 

 when recording II was made. This procedure was repeated on a subsequent day 

 to determine reproducibility. The later recordings are shown in B of this illus- 

 tration. 



The records show that at the 75° level (zero or reference level) channel 9 actu- 

 ally reads 76.0 units for the first day and 75.2 units for the second day. Two units 

 on the chart measure 0.1 °F. Thus, the chart records with no error on each day 

 but with a difference of 0.04° between the two days. Channel 7 reads 75.0 units 

 the first day and 74.8 high and 74.4 low on the second day, a minimum difference 

 of 0.01° and a maximum difference of 0.03° between the two days, and of 0.02° 

 in this particular run. 



At the 70° level, on the first day, channel 7 records a maximum 114.8 and a min- 

 imum 113.5 units, i. e., a difference of 0.06°F, and channel 9, 115.2 and 115.0 units, 

 a difference of 0.01°. On the second day, channel 7 reads 114.8 and 113.8, maxi- 

 mum and minimum, a difference of 1 unit or 0.05°, while channel 9 reads 115.2 

 and 115.0, 0.2 unit or 0.01° again. The maximum difference between the two 

 days' runs is no greater than this. Thus, the specified accuracy of ±0.1°F is well 

 attained. 



Actually, only one or two readings were at the extremes, the majority falling 

 within the limits of ±0.02°. 



The maintenance of an absolutely constant voltage to the bridge circuits is ex- 

 tremely important for accuracy. As an example, the indicating needle of the volt- 

 meter measuring this potential can be on the indicated mark but not quite cen- 

 tered on it, resulting in an error of 0.02°F. The voltage must be readjusted 

 whenever the operator changes any of the ranges. 



Although the multichannel recorder can be used to record sixteen separate 

 channels, it is often used for eight to four channels in which case the wiring is 

 rearranged so that one bead will actuate more than one channel. In this way, the 

 rate of recording each individual temperature is increased several fold. 



This unit has been used for ten years. Some temperature recordings have been 

 made in the Arctic and others in the tropics. It is sufficiently reliable to be left 

 unattended for hours, with an attached timing device marking the tape every 

 minute. Maintenance has been minimal and the entire system has proved highly 

 satisfactory. 



35 



