.334 Dr. J. Kerr's Experiments on the 



for the winged plate. This frame descends, without contact, 

 into a w r ater-tight cell of uniform thickness (about 4 centi- 

 metres), which stands in a fixed position, with its limiting 

 panes vertical. The winged plate, in a constant vertical 

 position, is immersed thus in the same vertical plate of air 

 and of water successively; and when all the pieces are 

 properly placed, the pencils of the refractor pass normally 

 through the composite plate, V traversing the strained pillar 

 of the winged plate, and H an unstrained wing. The strain 

 is impressed by a constant weight of 56 lbs., which is applied 

 by a sort of balance-beam, to a bolt that passes through the 

 iron traverse and acts on the sliding piece of the press. The 

 beam is raised and lowered by two assistants, who stand on 

 opposite sides of the table. Generally, the plate was in air 

 one half of the day, and in water the other ; and a good 

 many measurements were taken in both cases each day. 



As the range of measured effect was very small, the 

 probable error of observation was relatively great ; there were 

 also accidental disturbances, which could neither be wholly 

 prevented nor taken exactly into account. A good deal of 

 time was spent in attempts to get consistent readings. There 

 was always a clear decrease of effect from the case of air to 

 that of water ; but the decrease was very irregular, oscillating 

 about 30 per cent, through a considerable range. Even the 

 average of successive tens varied from about 25 per cent, to 

 36. Latterly, when all known precautions were taken, the 

 results agreed better, but never quite satisfactorily. I give 

 here the results of the last four sets, each of the numbers 

 being an average of about one hundred measurements : — 



Effect in air ='1632 -1712 -1712 '1712 

 Effect in water =*1U5 '1175 -1200 -1187 

 Fall per cent. = 317 31*3 29'9 30'7 



The smallness of values in the first column was understood to 

 be due to heterogeneity of strain. 



In a later set of measurements the method was different. 

 The plate was permanently immersed in water, and the strain 

 was produced by pressure of two screws on the sliding piece 

 of the press ; the effect in water was measured as in the 

 former experiments ; but the effect in air was inferred (by 

 prop. IV.) as the double of the effect given by the quartz 

 compensator, the latter effect being evidently the same in air 

 as in water. One advantage of the method was the com- 

 paratively large range of effect measured ; but along with 

 this there were some disadvantages. 



Several hundreds of measurements were taken in this way 



