I needn't have worried about the differences being too small to measure. They turned out to be absolutely huge.
The DL-103 is known for generating a lot of excess mechanical energy
that goes straight into the headshell. There's this big 7 kHz peak, dominating over the main armtube resonance at about 300 Hz, its harmonics, counterweight resonances etc. This is also approximately where the DL-103 has its 2-3 dB frequency response notch when measured with pink noise rather than a sine sweep.
Putting an aluminium cap over the plastic DL-103 body resulted in the resonant peak shifting down to about 3800 Hz, becoming slightly smaller but increasing in Q. Other effects included a smoother overall resonant picture, and lower counterweight resonances (although I suspect the reason for the latter is the change in counterweight position).