The use of the term ft, in contrast to /i, has not been entirely satisfactory. This is especially obvious in comparing the experimental deviations obtained by using Equations VI and VII. The possibility exists [as first pointed out to the authors by Kieselbach (8)] that the use of Ep0v0fi rather than Cv"f2 would prove more satisfactory in Equation VII. This follows from the fact that fi is a slowly varying function differing only slightly from unity at any time. The use of unity in place of fi, Equation VI, is, of course, very satisfactory. This explanation would constitute an argument in favor of the predominance of the E term. Some other factor might, however, be responsible for the deviations. The plate height contribution of the injection and detection devices cannot be held responsible, because the trend from larger to smaller values would be reversed, and this effect is as large as 10% only in extreme cases (as shown by measuring the peak width for short, blank columns, and extrapolating to zero length), and is usually much smaller. A great deal more experimental and theoretical work is needed on this subject.