change led us to repeat their work and to study the effect of added alkali on many exchanges. Like Winter, Carlton and Briscoe we were completely unable to confirm the effect in the case of sulfate, and are of the opinion that the observations were in error, or were due to exchange with silicate in the glass. We have some slight evidence of catalysis by alkali in the case of chlorate, bromate, iodate and arsenate, but the effects are so slight and the working range in the case of iodate so small (less than 2 $) that we cannot feel certain that the apparent effects are not due to silicate in the glass, or other experimental error. On the other hand, in the case of nitrate (run 24) perchlorate (run 42) selenate (run 58) and sulfate (runs 68, 69) we found no trace of such an effect of the glass container and a total absence of effect of the added alkali.Finally, in line with Winter, Carlton and Briscoe's experience, we found that alkali somewhat retarded the exchange of chromate, and we have further found that the presence of 0.2 N sodium hydroxide completely inhibits the otherwise complete exchange of sulfite and thiosulfate. Summary 1. Extensive density measurements by the method of the magnetically controlled swimmer have been carried out with a routine precision of 0.02 rd and an over-all reproducibility (including the purification of samples) of 0.1 rd.Attention is called to the advantage in comparing the densities of waters or dilute aqueous solutions, of working near the temperature of maximum density of the samples.Oxygen exchange between anions and water enriched in 01* was studied a t 95-100' in the case of 21 inorganic salts and CrOs.Resistance to exchange was found to increase toward the upper right-hand corner of the periodic table and with the number of oxygen atoms attached to a given central atom.Many of the exchanges appear to be promoted b y excess acid and retarded by excess base.Most of the salts studied exchange completely under the conditions used, several show no exchange at all, and only a few show partial exchange.
2.
3.
4.
5.6.
MADISON, WISCONSINSince its discovery2 the excess atomic weight of air oxygen (Dole effect) has been measured by numerous authors with fairly consistent result^.^ There has been critical discussion of the various methods used by the different authors but the chief uncertainties seem to be (1) the still imperfectly known degree to which different natural waters differ in density, ( 2 ) the possibility that samples even of the "same" water, such as "Lake Michigan," "Atlantic Ocean," or "Lake Mendota" used by different investigators or by the same investigators a t different times are significantly different in density, and (3) the possibility of different degrees of isotopic fractionation during the purification of the various samples, particularly as the same purification process produces a slightly (1) This work was supported in part by grants from the Wisconsin Alumni Research Foundation. (2) (a) Dole, THIS JOURNAL, SI, 2731 (1935); (b) J . Chcm. Phys., 4, 268 (1936...
