At 300 kV it should be done in water, at 2 cm depth, according to IAEA TRS 398. Preferably with a Farmer chamber.

Then you need to apply the appropriate attenuation factor corresponding to the depth of your tumour.

Some publication have come our since the TRS 398 suggesting that for x-ray beams at or near the surface, that measurements in depth are inaccurate because depth-dose curves for kilovoltage x-ray beams are difficult to measure and measurements on or near the surface will be less reliable using the "in-phantom" method.

Radu, do you think I should still employ a measurement at 2 cm depth?

Yes. I have done this and measured also the PDDs in a stationary water phantom, from 0,5 cm depth to 5 cm depth.

At the end of the day you need to work according to a standard, so te TRS398 what you have at hand. But, if you do research and you are not restrained to a standard, you can also do it differently: use a virtual water slab phanton, and you can work your way from the surface down to any depth you like (the slabs have widths between 1 mm and 1 cm, so you can measure the PDDs much more acurrately than in real water. You can take a look at the RW3 slab phantom from PTW, from instance.

Radu´s tips are the actual state. The change of X-ray spectra with depth in a low Z material like slab phantoms are for 300 kV and filtering with 2 mm Al less unsecure than some back scatter corrections. The 2 cm proposal of IAEA rep. 398 avoids low energy scattering electrons at the entrance of the flat paralel chamber which would falsify the measuring results, but dissapear very quickly out of the spectrum in the phantom and show no dose contribution in the therapeutic depth.

Although the above suggestions are opinions from people I presume have dosimetry expertise in some professional capacity, I in fact would only prefer a standardized dosimetry (appropriate phantom size and irradiation setting configuration) that can determine the dose calibration of at least a minimum of  98%  accuracy and above that. That can give assurance that you can attempt to determine which is more accurate for your requirement and be able to determine an actual  vis-a vis comparison of air vs. water dosimetry that you can rely. Then question of reliability and accuracy of such dosimetry has to be done by at least multiple repeat of such scheme to be statistically analyzable and verifiable.

@Vilmar,

you are right, multiple repeats help to verify your results, but such a technique does not guarantee that you have avoided systematic errors because each repeated result will be the same but still wrong.

Hanno Krieger's opinion makes the case that rad dosimetry professionals are incompetent of what they are trained to do on a regular basis.

Such opinion is obviously based from a naive comment based on non-familiarity of dosimetry calibration methodologies.