444 BELL SYSTEM TECHNICAL JOURNAL 



In the range where equation II is appUcable the relation is seen to 

 be a family of convergent lines with slopes (the constant A in this 

 equation) having values between 10 and 12. These convergent lines 

 focus at about 10 per cent M.C. (log per cent M.C. = 1).^^ The 

 actual value of the slope A in any test depends upon several factors. 

 It is primarily dependent upon the previous treatment of the cotton. 

 Water-boiled cotton which has been dried from the wet state at high 

 temperature in such a manner as to secure a high I.R. for a given 

 moisture content, in consequence, gives a line with maximum slope. 

 Exposure to high humidities, or saturation of the cotton with water 

 vapor causes the subsequent desorption and absorption equilibrium 

 values to lie on a line of less slope. In the case of raw cotton, the more 

 moisture absorbed by the cotton from a saturated atmosphere, the 

 lower is the desorption value of A ; its lower limit appears to depend 

 to some extent upon the time of exposure and the amount of moisture 

 absorbed. (Note the difference in the desorption slope after the first 

 and second exposure of the raw cotton to saturated air. After the first 

 cycle with 24.5 per cent maximum moisture content, A = 10.15; 

 after the second with 30 per cent M.C, A = 9.88.^^ This difference 

 is greater than experimental error.) 



Raw cotton shows a distinct difterence from water-boiled cotton in 

 one respect. On the second absorption cycle the slope A has a value 



content. This effect is not readily detectable, using the slow-period H.S. type 

 Leeds and isorthrop galvanometer. When first found, it was assumed that the 

 entire curvature of the curve above 10 per cent M.C. was due to this effect, but such 

 was not the case. The effect is not true polarization, but is simply due to electrical 

 heating. Above 90 per cent relative humidity for raw cotton and above 98 per cent 

 R.H. for washed cotton, the measuring current, using 100 volts potential is sufficient 

 to heat the cotton appreciably. This 1-R loss can raise the textile temperature about 

 0.1° C. at 90 per cent R.H., and about 10° C. at saturation for raw cotton. These 

 temperature rises were measured, using thermocouples of Xo. 40 wire braided into 

 the threads of textile mounted on the electrodes. 1 he heating effect causes evapora- 

 tion of moisture from the cotton, thus raising the insulation resistance. 



All measurements in this paper above 75 per cent R.H. for raw cotton and above 

 90 per cent R.H. for washed cotton were made with a special micro-ammeter having 

 a period of but 0.8 second, as compared with the period of the H.S. type galvanometer 

 of about 40 seconds. The temperature rise at saturation does not become evident 

 for at least three seconds after voltage application. Until this short interval has 

 elapsed the micro-ammeter gives a steady reading identical with the instantaneous 

 value, and as the thermocouple records increasing temperature the meter deflection 

 drops. 



1* This behavior is a hysteresis effect of a somewhat different character from that 

 observed in the two relative humidity relations previously discussed, since in this 

 case the effect is independent of relative humidity, and appears to be related to the 

 distribution of moisture in the cotton and to the manner in which this moisture is 

 held by the cellulose. This will be discussed somewhat more fully later. 



'^ The value of 24.5 per cent AI.C. does not necessarily indicate a true saturation 

 value, but only a M.C after exposure to a definite saturated atmosphere for one 

 hour. The 30 per cent value probably represents some value above the critical 

 saturation point at exactly 100 per cent R.H. (which would be e.xceedingly difficult 

 to obtain), since actual deposits of dew were visible on the sample. 



