860 SURVEY OF THE COAST 



From seven to twelve thermometers were laid constantly 

 over the scale and the standards, and were read at proper 

 intervals of time. 



The work bench itself was about double the length of the 

 scale. It was accurately adjusted before the work, and the 

 scale was so placed with respect to the windows of the room, 

 that the microscopes received their light from separate win- 

 dows. 



I'he bench was made of two planks three inches thick, 

 placed at right angles to each other, so that a transverse 

 section was in the form of the letter T. The top plank was 

 about twelve inches broad, and the whole rested on six legs. 



No fire was kept in the room while the comparisons were 

 making ; and for some time before, the windows were left 

 open day and night, to keep the different parts of the room 

 in an equal temperature, being that of the surrounding at- 

 mosphere. 



For the comparisons intended to be made in one day, 

 every thing was prepared the day before, and left in such a 

 state as to require as little handling as possible. This was 

 done, in order that the parts of the apparatus might acquire 

 throughout an equal temperature. 



All tliese precautions were necessary to obtain satisfac- 

 tory results, as is well known to men in the practice of such 

 operations. 



The probable error in the microscopic readings may be 

 considered as increasing with the number of these readings. 

 Having four metres, and the scale being sufficient to take in 

 two, I had the means of diminishing this eiror one half, by 

 comparing two at a time, instead of one. An equation be- 

 tween the results then enabled me to obtain the value of 

 each metre. This method had the advantage of removing 

 every prejudice from the mind of the oI)server in regard to 

 the readings, in as much as the combination of the diffe- 

 rent measures and the different influence of temperature 

 occasioned a variation which completely precluded previous 

 estimates. 



