The Graves Shield underneath the low beam filament covers a smaller part of the filament on H19 than on H4. So if you put an H19 in an H4 lamp, there will be no low beam -- significant amounts of unfocused light will be thrown upward, and there will be no horizontal cutoff line. If you put an H4 in an H19 lamp, a small, dim, and random distribution of light will result.
Optically speaking: the H4 shield, just like the R2 shield before it, was designed to create an asymmetrical shadow on a parabolic reflector; the projected edges of this shadow create the asymmetrical low beam. That was the only way to do the job when all we had was parabolic reflectors. Now we have highly complex reflectors that can create sharp cutoffs by precisely focusing filament images -- the edge of the filament itself is used to create the cutoff. That's how reflector headlamps that use single-filament bulbs (such as H1, H7, H11, HB4...) can produce a sharp cutoff without shadowing any part of the reflector -- in the old days, before complex-shape reflectors, such headlamps contained a shield that did the same thing as the shield built into the H4 bulb.
So the shield inside the H19 bulb serves only to reserve a portion of the reflector exclusively for the high beam. This way it is possible to optimize the low beam
and the high beam. Compare that to a design like H13, where it's difficult to optimize either beam because the whole optically-active area of the reflector and/or lens is used for both beams -- so it must be a compromise to do both jobs without ruining one or the other.
The shielded H19 is not the only way to do this, by the way. Quite a few years ago Ichikoh invented an unshielded 2-filament headlight system (bulb + reflector optics) producing optimized low and high beams -- you can read about it in
this technical paper. That system was not pursued for commercial production, for reasons that probably have nothing to do with how good it was (or would have been).
