Calorimetry has been recommended and performed in proton beams for some time, but never has graphite calorimetry been used as a reference dosimeter in clinical proton beams. Furthermore, only a few calorimetry measurements have been reported in ocular proton beams. In this paper we describe the construction and performance of a small-body portable graphite calorimeter for clinical low-energy proton beams. Perturbation correction factors for the gap effect, volume averaging effect, heat transfer phenomena and impurity effect are calculated and applied in a comparison with ionization chamber dosimetry following IAEA TRS-398. The ratio of absorbed dose to water obtained from the calorimeter measurements and from the ionization measurements varied between 0.983 and 1.019, depending on the beam type and the ionization chamber calibration modality. Standard uncertainties on these values varied between 1.9% and 2.5% including a substantial contribution from the kQ values in IAEA TRS-398. The (Wair/e)p values inferred from these measurements varied between 33.6 J C(-1) and 34.9 J C(-1) with similar standard uncertainties. A number of improvements for the small-body portable graphite calorimeter and the experimental set-up are suggested for potential reduction of the uncertainties.