The simplest form of disk-cutter consists of a amplifier, similar to that used to drive a loudspeaker, connected to a cutting-head having a stylus connected to a coil, which is placed in the field from a strong magnet (or, more usually in later designs, a magnet within a coil). When the signal is applied to the coil, the stylus moves and engraves a groove in the blank disk. (There is of course a lot more to it than that, but we are considering only the basics here.)
However, because the cutter head's movements translate the amplitude swings of the original signal into velocity - the rate at which the stylus moves during its swings - low-frequency signals would be recorded with a much larger swing than high-frequency signals of the same original amplitude. In order to keep the movements of the groove much the same at all frequencies (given equal level signals) it is necessary to use a circuit to introduce - in the theoretical situation - a 6dB/octave cut as the frequency decreases - i.e., halve the frequency and you halve the voltage.
In the reverse situation, that of a reproduction head, the principal is that of a wire moved in a magnetic field - it is the rate of cutting 'lines of force' that matters. The cutter head works exactly in reverse, like a simple motor, where increased voltage means increased speed. Therefore the constant amplitude groove theoretically achieved produces a signal where the bass is low and the treble high: so a 6db/octave cut with increasing frequency would be called for.
In the real world, losses in the head with increased frequency complicate the issue. Early cutter heads were highly inefficient, and so, while the bass cut described above was used, the treble tailed away, resulting in equal groove modulations (movements) up to mid frequencies, but decreasing above that.
To compensate for this, the playback characteristic boosted the bass below 200 Hz but left it flat above that - effectively providing a 6dB/octave boost to the higher frequencies (and the surface noise). With the later improvements in cutters, it was possible to pack more treble onto the records, and so new equalizations provided for a 6dB/octave cut above a turnover frequency which varied between 3.4 and about 6kHz depending on the system. This meant that the surface noise became less obtrusive. It was also common to flatten out the bass at the very lowest frequencies to reduce the boost of rumble from the turntable.
Similar techniques were applied to microgroove records, and the final standard, RIAA (CCIR in Europe is the same) provides for a bass boost below 500.5 Hz and a treble cut above the lower frequency of 2.2125kHz - the latter reflecting the considerably increased amount of treble which can be cut onto an LP.
It can therefore be seen that playing a 78 with RIAA equalization - all that is available to many people - produces far less top than is correct -particularly for the earliest electrical, where the result is akin to turning the treble control right down. (Turning it right up gives an improvement, but doesn't touch the important mid-range.)
Use of the correct curves when reproducing 78s produces a startling improvement in the quality (although admittedly the surface noise can become a problem): many of these recordings are much higher quality than you might suppose.