200 SCIENCE PROGRESS 



problem then is equivalent to that of finding the true vertical 

 in a moving aeroplane. This is rendered difficult by the non- 

 uniform motion of the aeroplane, which alters the apparent 

 direction of gravity. Prof. Russell points out that, if a devia- 

 tion of 15' is adopted as the greatest allowable, the speed of the 

 aeroplane must never be permitted to change at a rate exceed- 

 ing one mile per hour in ten seconds, nor must the radius of 

 curvature of the path be less than ten miles (assuming a speed 

 of sixty miles per hour). This requires careful piloting, but 

 is possible with a good pilot who is accustomed to his machine. 

 The first apparatus used was in the form of an artificial horizon, 

 and consisted of a pendulum mounted in gimbals so as to 

 swing with freedom in any vertical plane, and bearing at its 

 upper end a mirror adjustable so that its reflecting surface is 

 horizontal when the pendulum swings freely. The lower end 

 of the pendulum swung in viscous liquid to damp its motion, 

 and the whole instrument was enclosed in a case with a glass 

 cover, for protection from wind. The observations with such 

 an instrument are similar to those with a mercury horizon, 

 and, in spite of small, rapid vibrations, exhibit no difficulty 

 after a little practice. This instrument was found to give 

 satisfactory results, provided the machine was not accelerated, 

 but suffered from the disadvantages of being heavy, difficult 

 to move, and at times inconvenient for observation. 



It was replaced by a bubble sextant, which is an attach- 

 ment to the telescope of an ordinary sextant, containing a 

 level bubble which is reflected into the field of view of the 

 telescope. Any object which appears to be superposed on the 

 centre of the bubble will actually be on the true horizon. The 

 manipulation of this instrument is similar to that of a sextant : 

 it is not affected by engine vibration, and wind does not usually 

 offer any serious inconvenience. Accelerations of the motion 

 of the aeroplane are not so conspicuous as with the artificial 

 horizon, and with good piloting results of sufficient accuracy 

 are easily obtained, by taking the mean of several consecutive 

 readings. 



Prof. Russell gives particulars of the reduction of the 

 observations, which follows St. Hilaire's well-known method. 

 He states that the average time required to secure a set of 

 ten observations was 4*5 minutes, three minutes were required 

 for computation, and two minutes for plotting the Sumner 



