OILS, ETC., USED IN THE PAINTING OF PICTURES. 467 



twenty days the distance between points was 0.57 millimeter. On 

 now putting into saturated water vapor, the distance between the 

 points became 0.36 millimeter instead of 0.28 millimeter, showing 

 a distinct and definite contraction on the original canvas. 



A coat of umber was now laid on the ocher and allowed to dry 

 in saturated air. In three days the umber was dry, and the distance 

 between points was 0.34 millimeter. On noAv putting back in dry 

 air a further slight contraction took place, the distance between the 

 points becoming 0.66 millimeter. 



A coat of yellow ocher was now put on and left over sulphuric 

 acid ; at the end of seven days the reading was 0.73 millimeter, and 

 at the end of a fortnight 0.72. 



These figures are a little difficult to follow, but leaving out small 

 fluctuations they bring out the following facts very clearly : 



In the first place, the total expansion, from dry to moist air, of 

 the canvas tested, was 0.28 

 millimeter for 2 centimeters 

 length, or for 1 centimeter 

 (that is, 10 millimeters) it was 

 1.4 per cent. In the second 

 place, the drying of the thin 

 layers of paint produced a 

 total contraction of the canvas 

 amounting to about 0.16 milli- 

 meter. This contraction did 

 not necessarily show itself 

 while the paint was drying ^^^- 1— apparatus for measurixg expansion and 



, n -I , 1 • ' ^ Contraction of Canvas. 



under fixed atmosplieric con- 

 ditions, but as soon as the canvas was set moving by change in atmos- 

 pheric conditions, it asserted itself. 



It is of interest to compare the magnitude of these movements with 

 those required to produce a badly cracked picture. 



The cracks in the picture mentioned above varied in diameter from 

 0.12 to 0.3 millimeter, roughly averaging 0.2 millimeter, and measure- 

 ments in different directions showed an average of about twelve cracks 

 to 10 centimeters =0.24 millimeter per centimeter, or about double 

 the total expansion of the canvas, as tested. 



After nine months the strip of canvas painted as above described 

 was alternately put over sulphuric acid and over water some two 

 or three times, expanding and contracting freely, but without crack- 

 ing the paint. Another possible source of mischief had to be iuA'^es- 

 tigated, and that was the freezing of a damp canvas. Two pictures 

 were taken, one a canvas which had been primed tAvelve months before 

 with sulphate of lead and zinc oxide, the other a portion of canvas 



