of a Charged Franklin's Plate. 365 



those pure and intense effects which I obtained long ago in 

 liquid dielectrics, particularly in carbon disulphide, and in a 

 sensibly uniform field. The strains assumed in the first and 

 third statements are undoubtedly real ; and it must be left to 

 experiment to determine whether, and how far, they take 

 effect in any actual case. With regard to my own early 

 experiments on thick plates of glass, I am not aware of any 

 evidence to show that these very small strains contributed to 

 the observed effects in any degree whatever ; but I pass from 

 these points to our proper subject. 



The second of the preceding statements is not true in any 

 other sense than as a record of several inadequate experi- 

 ments. I find that a charged Franklin's plate, with uncoated 

 margin so wide only as to give a small, working difference 

 of potentials, acts upon transmitted light in the same way as 

 olive oil, amyl oxide, or any other negative dielectric. .My 

 earlier experiments on the Franklin's plate need not be de- 

 scribed, as they were chiefly of use in deciding between 

 methods, and in showing the difficulties that had to be avoided 

 or overcome. My last set of experiments on the subject were 

 finished in July 1882 ; and the following account of them is 

 drawn from full notes that were taken at the time. 



One of my best plates was a rectangular piece of good 

 English plate-glass, exactly § of an inch thick, 2| inches 

 wide, 4| long. This piece was one of a set, which were cut 

 out of separate plates or fragments, and were then put together, 

 and worked in block as one thick plate. The smaller ends of 

 the thick plate were ground to sensibly parallel planes, and 

 were polished carefully, till the glass gave a perfectly clear 

 and undistorted view of neighbouring objects through the 

 length of the block. Before going further with the prepara- 

 tion of the plates, I tested them separately in the polariscope; 

 for I had found already that, unless the plates were tested 

 and carefully selected, the experiments in view were almost 

 hopeless. 



Along with a couple of Nicolas prisms, two other pieces 

 were employed — a rectangular slip of thin plate-glass used 

 as a hand-compensator, and a larger slip used as a fixed 

 compensator, the latter being placed in a simple screw- 

 press, and subjected to horizontal tensions and compres- 

 sions, which could be made to vary continuously through a 

 considerable range. The two Nicols were placed in hori- 

 zontal line with a good paraffin flame, and were exactly 

 crossed, their principal sections being at 45° to the vertical. 

 The plate was then fixed between the two Nicols, its faces 

 \ertical, and its terminal polished surfaces perpendicular to 



