The measured penetration depth, l, of deuterium pellets injected into the Joint European Torus (JET) confirms some features of the neutral gas shielding (NGS) model, but not others. The scaling of l with plasma and pellet parameters agrees with the NGS model, as in earlier ASDEX studies. Pellet velocity was varied over the range 0.46-1.35km/s in the JET experiments to test specifically the scaling of l with velocity. This scaling also agrees with the NGS model. However, the penetration is deeper in JET than in ASDEX when corrected for expected machine size dependence. Furthermore, the measured penetration depths are greater (by nearly a factor of two) in JET than those predicted by local ablation calculations using the neutral gas shielding model with an incident Maxwellian distribution of electrons. Plasma shielding used in previous modeling of the JET penetration data can account for the additional shielding but also removes the observed velocity dependence. The implications of both the scaling observations and the penetration depths for improvements in ablation theory and models are discussed.