The most recent experimental data on the specific heat of hydrogen are reviewed and average values are given for various temperatures up to 1000\ifmmode^\circ\else\textdegree\fi{}K. At room temperatures many values are available but they do not check among themselves. At high temperatures the values of the specific heat are not at all certain. Computations are made using different minimum values of the rotational quantum numbers and different a priori probabilities for both rigid and elastic models. The most successful temperature specific heat curve at both high and low temperatures and based upon an elastic model is obtained with (a) the a priori probability $p=2m$ for whole quanta and (b) $p=2m$, $m=\frac{1}{2}$ excluded, for half quanta. It is shown that half integral vibrational quantum numbers do not appreciably affect the specific heat except to change an arbitrary constant. In scheme (b) the normal vibration frequency agrees almost exactly with the value which Witmer has obtained from spectroscopic measurements.