M values are given in units of pressure. So for example the M value for the 4 minute compartment of nitrogen is 106.4 fsw in Buhlmann's ZHL16(1990) table. In this case fsw (feet of sea water) is used as a pressure unit. Pressure gradients are also pressure because they are just the difference between compartment pressure and ambient. Surfacing M values (M-naughts) of course are also in units of pressure. If you wanted to you could work in atmospheres, bar pascals, fsw, msw or whatever as long as you stay consistent. Standard practice has been to use fsw or msw which makes relating pressures during the dive easier.
So for example: M (fsw)(the maximum allowed pressure gradient)=dM(unitless)*depth(fsw)+M-naught(fsw) -- everything has units of pressure, except for deltaM which is a slope and is unitless.
The confusion may be that Haldane used pressure ratios in his work around 1908. Workman who originated M values, around 1965, used differences.