Have a couple comments peripherally related to BSA ports and Jagbruno'ss observation about historical port shapes. But before that want to jump back to an earlier part of the discussion relating to HP.
We can debate ad nausuem exactly what HP a late style XR 750 makes, i.e., who's dyno, who's calibration, rear wheel, crankshaft, etc. That said, I believe it is well accepted that they make somewhere on the order of 100 RWHP give or take. You're definitely not keeping up with an oval port XR on an 85 HP bike. Perhaps the number is 95 HP, or maybe it's 105 HP, but it’s somewhere in that neighborhood. I have not seen unrestricted oval port XRs on a dyno, but I have seen 32 mm restrictor plate engines on a dyno and they made on the order of 85 RWHP. Thus it does not seem a stretch or at all unreasonable to think the unrestricted version could produce somewhere in the neighborhood of 100 HP.
Over the years other marques have made considerably higher power than an XR, but just because a bike makes more power does not imply that that power can be put to the ground efficiently on a flat track or that it can keep up with an XR. Harley stumbled into something quite unique and extremely good for flat track racing with the KR/XR engine configuration (v-twin angle, flywheel mass, flywheel width, power pulse frequency, etc - although recently [last year or 2] AMA made it illegal to twingle them - evidently the race audience hated the sound!). The 100 HP number also comes up frequently, not just for XRs, but for the stroked Kawasaki vertical twins that compete against them and make similar if not more HP. Kawas won all the GNC mile events this year, so they're definitely making some serious power, and as we've noted on the forum in the past, the Kawas epitomize what modern day vertical twin performance can be).
But getting back on track, since considerable content of the present discussion is about port design, and focused on the virtues of the XR oval port head, we tend to overlook the fact that for 15 years preceding the introduction of the oval port, the Al headed XR existed and won myriad races with a round port head, and as has been pointed out previously, it was basically a BSA Goldstar port. Reviewing the past might allow us to see the present more clearly, so to that end am offering a bit of perspective, via the images below, regarding the XR round port that preceded the oval port design of Ken Augustine.
Below are images of castings taken from the late style oval and early style round port heads. In this case, the round port was one done by the king himself, CR Axtell, and from Axtell's numbering system indicates it was done circa 1984 (perhaps lcrken is familiar with the numbering system and can confirm the date code). Flow testing the round and oval port heads fit with radiused entry at 10" WC gives flows of ~ 125 and 140 cfm respectively.
Harley XR round port head ported by CR Axtell
It's interesting to note that the top views of the 2 ports make the late style oval port look gigantic, but in the side view the late style port is shallow and svelte in comparison to the round port. As others have noted previously, the shallow port (top to bottom) allows the floor to be elevated, thereby creating a larger radius curve into the throat area, while simultaneously minimizing the differential in radius between the short and long sides of the port, thereby minimizing velocity differentials between short and long sides as well.
Top view - oval port right
Side view - oval port bottom
Round port - IN left
Oval port - IN left
I have not measured the volume of the oval port along its length, but my intuition is that the x-sectional area of the oval port, even though it looks large and bulbous where compensation is made for the extra volume of the guide and valve stem, is relatively constant along the port's length. The port x-sectional area along its length is something that should be easily assessed by anybody having the SolidWorks database of the oval XR port. Anyone have the database, and if so, could they provide a sense of the x-sectional areas along the length of the port, i.e., from upstream of the guide all the way to the throat?
The images below of the port entry into the head casting show that the elevation of the round and oval ports are essentially identical, thus as described above it is the oval nature of the port that enhances flow by elevating the floor, increasing the short side radius and minimizing the differential between short and long side radii.