Exactly what I was wondering. These sorts of studies usually make assumptions about the rate of mutation across time in order to estimate the date of various divergences or of the ur-form, but these rates are difficult to calibrate. For a cautionary tale, see also glottochronology.