Attempts to use the decay of radio signals reflected from individual meteor trails to study the upper atmosphere have revealed a puzzling inconsistency in the signal behavior. An earlier paper pointed out that this inconsistency remained even when the previously postulated sources of error were eliminated. As a result, an irregularly ionized trail model was proposed and shown, by calculation of signal characteristics, to be capable of accounting for the observations. This paper presents results of a new experiment which permitted the determination of the ionization profiles as the meteor trails were formed. The predicted irregularities were found, even for trails which exhibited apparently "ideal" underdense signal characteristics.
