12 Ravletgh, Tlie jSIecJimiical Prvidples of FligJit. 



provided for estimating the couple needed to prevent 

 rotation, \vc should obtain the efficiency of the vanes at 

 the given obliquity and speed. Observations at the same 

 speed and at other obliquities would then give the means 

 of determining the law of obliquity. 



Such a proccedure would be analogous to that 

 adopted in former experiments with whirling machines. 

 The essential feature of the present method consists in 

 setting some of the vanes to compensate others inclined 

 at different angle. The balance of effects is independent of 

 the speed of the wind, so long as it is uniform over the whole 

 section in operation. To guard against errors that might 

 arise from a deficient fulfilment of this condition, I have 

 preferred so to arrange that opposite vanes were inclined 

 always at the same angle. For example, two pairs of 

 op{:)osite vanes might be set so that their planes make an 

 angle of 6'' with the axis. The remaining pair of opposite 

 vanes would then be set at a greater angle, and this would 

 be varied until no tendency remained to turn in either 

 direction. The exact point of balance could be inferred 

 either from the absence of observable effect, or by inter- 

 polation from equal slight effects in opposite directions. 



As has been suggested, the motion itself ma\- be either 

 horizontal or vertical. Fair results may be obtained 

 indoors at a walking speed, and my first idea was to 

 determine balances by holding the wheel overhead while 

 travelling in a dog-cart at lo or 12 miles per hour. lUit 

 wdien the axis is horizontal, much time is lost owing to the 

 necessity of readjusting the centre of gravity after almost 

 every shifting of the vanes. With a nearly vertical motion 

 the position of the centre of gravity is of less consequence, 

 and it was found that very good results could be arrived at 

 by somewhat rapidly lowering the apparatus while held in 

 the hands with axis vertical. It is po.ssible that part of 



