The determination of a primary standard of kerma in air is generally based on the use of an absolute cavity ionization chamber, with graphite wall and filled with air. For laboratories in possession of primary standard graphite calorimeters, it appears possible to derive by a transfer method air kerma from absorbed dose in graphite. This derivation can be achieved using a transfer instrument, e.g. a small air-filled cavity chamber, successively irradiated in the graphite phantom at the reference depth and in free air. The advantages of this method are the replacement on an absolute current measurement by a relative one, the replacement of the graphite-to-air stopping power ratio by the quotient of that stopping power ratio in free air by that in graphite, the elimination of the effective volume of the chamber and of W, the mean energy expended per ion pair formed in air. In return, a correction factor for the perturbation of the particle fluence, due to the air cavity irradiated in the graphite phantom must be applied. A critical analysis of accuracies proper to both methods is made, the expected gains in metrological safety with the new method pointed out, and preliminary results are presented. (author).