ATMOSPHERIC IONS. 151 



to be found in the galvanoscope. Changes of temperature produce vorti- 

 cal currents in the capsule which modify the deflections of the aluminum 

 foils. In the mean ranges, one scale part of double deflection is equivalent 

 to about 6 volts, or to i ,000 ions per cubic centimeter. Therefore, the 

 presence of any secondary disturbance like the one in question is of very 

 serious consequence. Prior to measurement, the apparatus was naturally 

 left in the cold air out of doors until temperature uniformity was pre- 

 sumable; but this is not sufficient, as the special observations in the 

 early part of February show, even for a galvanoscope dried with sodium. 



Turning first to the leakages due to conduction, etc. (v in the above 

 equation), direct experiments made at different times showed values of 

 0.073, 0.067, 0.120 volt per minute, or less than 0.9 volt for the ten 

 minutes of observation. This is equivalent to an excess of 150 ions per 

 cubic centimeter. As it is applied equally to the positive and to the 

 negative ions, is independent of the size of the deflections, and the same 

 no matter whether the deflections on both sides are equal or not, it has 

 no bearing on the outstanding errors in question. It was not deducted 

 from the ionizations (n) of table 53, which are therefore slightly too large. 



Trials made between February 4 and 1 1 showed that in almost every 

 case the first measurements (whether for positive or negative ions) , even 

 after the galvanoscope had been exposed to the cold air for some time, are 

 too large. This discrepancy may at times extend to the second and third 

 observations (February 6, 9, 11). Thus on February 9 the estimated 

 positive ionization would be a thousand and zero, for instance. Usually, 

 however, the second and third observations are liable to be trustworthy 

 (February 6, 10, 11, etc.). Hence electroscopic apparatus which can not 

 be left permanently out of doors, but is taken from a warm room into 

 the cold atmosphere, even if it is sodium dried, is not liable to show war- 

 rantable results after mere waiting for uniform temperature. It seems 

 additionally necessary to pass a large volume of cold air through the 

 condenser, or to make successive measurements in series. The observer 

 is usually in doubt, whenever the positive and negative ionizations differ 

 widely, so that at least three tests must be made. The tendency of the 

 apparatus to show spurious results is usually indicated by an inequality 

 of deflections of the foils on either side of the vertical. They may increase 

 to a maximum after charging and then decrease regularly. The latter 

 probably finds an explanation in the gradual cessation of a down-pouring 

 cold air current near the sides of the capsule of the galvanoscope, but the 

 persistence of unequal deflections must follow from other causes. Re- 

 membering that the air is desiccated internally with metallic sodium, it 

 seems hardly creditable that there can be a precipitation of moisture 

 from this dried air on the aluminum foils; and yet the behavior is such 



