That rolloff is typical of directional microphones designed for placement close to their sound sources. Neumann typically shows on-axis response at a 1-meter pickup distance. Other manufacturers may use shorter measurement distances (30 or even 15 cm).
The reason this matters so much is proximity effect. At 15 or 30 cm (ca. 6" - 12") it boosts the bass and midbass significantly across a wide range of frequencies. So if you plot a graph for such close placement (as DPA does, and Audio-Technica for some graphs), the graph may show full low frequency response--but at greater miking distances, that bass won't be there in nearly the amounts shown. It's not a small effect--it can very well be 10-12 dB of bass boost close up. If that boost is what creates flat low-frequency response at close range, you need to deduct that amount to understand what you will get for more distant pickup.
Consider the two graphs that I've attached, from two older, closely-related Neumann microphones (identical amplifiers; different capsules based on the same acoustical design). These curves were both made at a 1-meter distance (OK, actually they were almost certainly taken from farther away, then adjusted to approximate a 1-meter response--and obviously smoothed for presentation; all of that is typical practice). The point is that if you measured the KM 85 from 15 cm, where there's much more proximity effect than at 1 meter, its curve might look almost exactly like the KM 84's 1-meter curve. Then if you went by the curves and didn't consider measuring distances, you might buy the "speech cardioid" KM 85 without realizing that it was a microphone for close pickup, rather than a general-purpose studio cardioid.
Often we as consumers don't know how the manufacturer's curves were obtained. Even if we do know, how exactly does one convert a 15- or 30-cm curve into a 1-meter curve? There's no fixed formula. And even at one meter there is still some proximity effect. It's an industry-wide problem but it persists because the manufacturers don't seem interested in solving it--or more precisely, no one wants to be the first to show realistic performance curves at various distances, and thus risk losing customers to the other manufacturers who reveal less, and let consumers keep their illusions. When a manufacturer says that their microphone was designed mainly for a certain type of use, we should understand that it probably isn't equally good for other situations that have very different technical requirements.
(Pressure transducers don't have proximity effect--single-diaphragm omnidirectional microphones--but all directional microphones do have it to some extent.)
--WiFiJeff, the article seems to say that smaller diameters such as 8 mm were considered, but 12 mm was ultimately chosen. That dimension normally refers to the active area of the diaphragm (i.e. the part of the Mylar that gold is evaporated onto) rather than its entire physical size. 12 mm is only slightly smaller than average as the diameter of the active area for small-diaphragm condenser microphones.
--best regards