What's with 4 ohm speakers?

Sensitivity.

Take a typical "8 ohm" midwoofer and double them up in parallel and you get 6 dB more. Realistically, more like 3.5 to 4 dB but also drops impedance by half. Problem is that impedance varies with frequency and the same midwoofer will have a minimum impedance, usually around 200 Hz. Some more than others.

The last drivers I used in a project had a minimum impedance of 5.6 ohms. In a MTM that's 2.8 ohms at a frequency that demands current. A 4 ohm tweeter doesn't have that current demand. Found a trick to eliminate that impedance dip but it doubles the cost.

Subwoofers are often 4 ohms because the trade-off for bass extension is sensitivity or size. So, they often use lower impedance to get more sensitivity from smaller drivers in smaller cabinets. Automotive subs are sometimes only 2 ohms. Same applies to woofers. Do you want bass? Do you want reasonably sized cabinets?

There's also the advantage of using smaller value inductors with lower impedance and less resulting phase shift.

Tell you what I'm going to do. Got another couple projects coming up this summer. I'll make up a 'quick and dirty' mono cabinet with 4 small, induvidually sealed midwoofers. Each will have it's own terminals that I'll hook up individual, parallel pair, series pair and series/parallel, measuring amperage, voltage and frequency response. Don't have state-of-the-art testing gear but should be consistent. No crossovers. Bottom woofer will be at least 1.5 feet above floor. Planned drivers are Dayton DS135-8's, which are fairly conventional except for small size and required cabinet volume.

Since the question was never about percieved sound quality or matching amps, (tube or SS) I'll just use what I have.

Never actually seen this done before, just math and theory.

Speaking of tweeters, Morel offers nearly identical MDT29's in both 4 and 8 ohm versions. As rated, the 4 ohm is 92 dB sensitivity and the 8 ohm is 89 dB. For a MTM, the 4 ohm would be more likely.

So, just by halving impedance and not doubling driver area and motors, you get +3 dB.

Finally got around to that test I mentioned earlier.

So I built a single test speaker with 5 individually sealed Dayton DS135-8's (5"). Not normally a choice for a sealed enclosure but fairly conventional except for a high Qms,low volume requirement and low price. These are nominal 8 ohm rated drivers. They were arranged vertically with 9 inches center to center and a 9 inch wide baffle. That approximates a typical MTM setup but without the T. I kept the volume and the microphone fixed. This is an Omnimic USB microphone. I tested with both "pseudo-noise" and short sine sweeps and measured max SPL. The amp used was a bridged Plinius SA-100.

For a starting point, I measured a single driver at 78 dB C weighted. (+- 0.1 dB for either test)

Two in parallel (4 ohms) measured +4 dB higher than a single. A touch lower on the short sine sweep.

Two in series (16 ohms) measured -1.5 dB lower than a single.

Four in parallel/series (8 ohms) measured +1.7 dB higher than a single. In earlier tests with this setup, I did notice a comb filtering effect and a dip around 1 KHz at the same mic location.

Won't draw any conclusions but I encourage anybody else to try a similar experiment. Results do go against standard convention. With closer spacing, or other factors, results may vary.

This test speaker was for more than just this. It also proved that I would need five of these drivers in a parallel/series 2.5-way setup to achieve some BSC and a fairly flat in-room response.