Moving at the speed of a thousand racing snails today i finally got to see a result of some time well spent reseaching and talking to people.. I formed a plan, executed it and seem to get what i planned.

Bevel heads have a straight shot port. Condemed around the world of performance because basically they are a shit angle to use. One of the cosworth guys worked out that 15-18° off valve angle is where the action is. One of the approaches is to go to 60° valve angle instead of 80°. Expensive and a lot of mucking around. So many tackle the port with varying degrees of success. So i thought i'd give it an in the shed go.
To get torque you need an increase in dynamic cylinder pressure. In other words stuff more air in. Th" big ports are us" crowd from way back reined for a while, but if you go too far air slows down. The swirl team threw in their 2 cents worth with good success. Team "swirl" belives better burn is the secret of life. But racey boys wanted more revs and more power, so team swirl took a back seat. Glue was used, but technically it was very good back then, and much heart ache was felt.
Time moved on and the bevel passed into history.

So my target, get more air into the cylinder and get it swirling. (Not too fast or the fuel gets flung out of the charge). Now as a young tacker i'd pulled out the ports by 1 1/2mm, which gave the engine a bit more go, but nothing really impressive. This time i've used velocity porting. (Add glue to make the choke smaller and thus causing higher velocity which has an effect of increasing the inertia of the mass of air in the intake track. Result is increased charge in the cylinder. Secondly with glue you can change the angle of attack of the air on the short side of the port. This pulls the boundry air closer to the short side, and thus increasing the working area of the seat throat, allowing more charge to get in. Thirdly with glue the bias of the air flow, which effects swirl. Combining glue and bowl shaping, it is possible to direct the air on to combustion chamber just about anywhere you want, this changes the shape of the spiral, either deep or shallow. Combined with more short side air running down the bore and more velocity i estimate a much better mix in the cylinder.
Using the mighty $5 vacuum cleaner flow bench i had a pressure figure. Pressure and velocity are conected to each other. As flow increases, pressure decreases. As with a dyno, what counts is taking a reading, change something and take another reading and note the direction and size of change.
My early moded heads lifted 37mm of water at 25". A standard 860 Gt Springer head, lifts 40mm. The new ports lift 45mm of water.

Then the cool part. String. I layed string at various locations aroung the intake. The Gt port made the string sit about 2-3 mm off the wall around any part of the port wall.
The string on my one one side behaved similar to the standard port, string lifted a couple of mm's just past the guide and into the bowl on the side wall, but the opposite side hugged the wall through the bend and into the valve seat throat. On the short side on the floor of the port the gt lifted slightly, on the glue section of the new D shape it hugged the floor. The string was held at 9 o'clock, 10, 12, 2, 3, 5,6 and 7 o'clock. At 12 oclock the string showed what happens arounf the valve guide and into the bowl. By adding bowling where before there was sod all the string lay closer to the valve seat than on the gt and had a beautiful curve in it. The bowling seems to act like a self adjusting cushion maintaining this curve. I sped up the air by sticking my finger into the port. All very cool stuff.

I also finally found an MIT paper with a formula for working out sonic reversion waves in an intake track for motorcycles. It gives you an rpm for a given lenght and diameter of intake at which a supercharging effect happens. With a 52mm velocity stack on a 40mm id over 280mm total, the rpm is 6950. Peak torque on a twin bevel is about 6500. So the correct velocity stack has the effect of increasing the duration of peak torque. Or you could run a shorter stack and increase the value of peak torque. Once again cool stuff.