The melting curve of aluminum was measured in a laser-heated diamond cell up to a pressure of 0.8 Mbar in order to test the agreement between this technique and shock wave measurements, which has been lacking in the case for iron. At this pressure, which is over an order of magnitude higher than in previous experiments [1, 2], the melting temperature is 3800 K, comparable to that measured for iron at 2 Mbar . The present results for aluminum extrapolate smoothly to the previous melting measurements in a multi-anvil apparatus to 60 kbar and to the calculated shock melting point of 4750 K at 1.25 Mbar. They are also in excellent agreement with theoretical calculations. A review of the shock data reported for Al, Ta and Mo, close-packed metals, in which a break in the sound velocity-pressure curve is used to determine the melting pressure, shows that the change in velocity at melting is about 10% for all three metals. In the case of iron, the sound velocity data have been used to infer two transitions: a solid-solid transition at 2.0 Mbar and melting at 2.4 Mbar, each of these transitions having about a 5% change in sound velocity. It is unlikely that a phase transition between close-packed cubic structures will have a 5% velocity change, the same as is found in the melting transition. We therefore suggest that for iron there exists only a single transition, starting at 2.0 Mbar, a region of incomplete shock melting between 2 and 2.4 Mbar, and a total change in sound velocity of about 10%, which is closer to the value of the other metals studied. This interpretation introduces a very good agreement between the shock melting results of Brown and McQueen  and diamond cell measurements for iron  which has up to now been lacking.