EDWARDS. — MANOMETER DEVICE FOR AIR THERMOMETERS. 645 



prisms is unchanged by snuiU values of [i and y. This can be ilbistrafed 

 by the following table of values of x and a:,, and the errors introduced 

 when the mean is taken. 



= 30° u, = 1.0930 



If rf = .2 mm. and ^ = y =^ 10' the error in using the mean of x and x^ 

 is 0.0001 mm. 



If the plate is out of the horizontal by having been rotaited around a 

 horizontal axis parallel to the scales, the reading of one edge line will be 

 increased and the other decreased to the same extent, so that the error 

 is eliminated in the mean. 



The light coming by reflection from the rear prism H has a somewhat 

 longer path than the light coming from the front prism ; therefore to 

 have all the images in focus at once, a block of glass of requisite length, 

 in this case 2.1 cm., is inserted in the path of the light from the rear 

 prism. It is not shown in the drawing. 



The advantages of the above procedure may be summed up as follows: 

 The images observed cannot be displaced and distorted, as the path of the 

 light lies entirely in air and optically worked glass. All errors of level- 

 ling are eliminated. When used in a constant-volume gas thermometer 

 the quantity of gas between the lower meniscus and the thermometer 

 bulb is reduced very much. The breadth of the beam of light coming 

 to the telescope is unrestricted, so that the resolving power is not cut 

 down. The meniscus does not have to be brought to a definite point, 

 but only near it, the remaining distance then being read in the cathetom- 

 eter telescope. This is especially convenient in barometer work where 

 two surfaces are to be simultaneously read, and the setting of one alters 

 the other. 



Jefferson Physical Lahoratort, 



Harvard University, 



Cambridge, Mass. 



VOL. XL. — 35 



