Through 1971, Yamaha built its 250cc and 350cc two-stroke twins on completely different bottom ends—the 250 vertically split, the 350 horizontally. In 1972, it did the sensible thing, building both twins on a common modular crankcase with a 54mm stroke. Both inherited the two pairs of crankcase-fed transfer ports pioneered in the 1969 250cc TD2 and abandoned the use of "afterthought" slot ports fed through holes in the pistons. Transfer ports continued to look a lot like those of TD2, with the "A" pair flanking the single oval exhaust port having flat roofs and considerable backsweep toward the non-exhaust cylinder wall. The "B" pair continued to be much smaller than the "As" and were aimed at each other rather than at the non-exhaust cylinder wall.

When Honda began to make two-stroke dirt bikes (whose engines would in 1982 become the basis of its first two-stroke Grand Prix bike, Freddie Spencer's NS500 triple) their "A" and "B" transfers looked quite Yamaha-like. Instead of Yamaha's signature single oval exhaust port, however, Honda adopted a T-shaped exhaust port with a center divider (despite the need for fussy piston modifications to keep it lubricated). The cross of the T overhung the "A" transfers, and such a wide exhaust port could therefore be effective even when opening a bit later than the single Yamaha exhaust port. The later the exhaust port opens, the more expanded and cooled the combustion gas becomes. When it does rush out as the port opens, the gas heats the piston crown less. This potential for a lower piston temperature can allow knock-free use of a slightly higher compression ratio, which increases torque.

Such a port divider, being exposed on all sides to the outrush of exhaust gas, runs hot. As the piston clatters up and down its bore, its rocking can cause the divider to hammer softened piston material into the piston ring groove, trapping the ring and destroying its seal. To lubricate the divider, small holes were drilled in the exhaust skirt of the piston along the track of the divider. This allowed crankcase compression to push oily air onto it. To reduce pressure on the divider, it was also common for builders to relieve the surface of its hard plating (or liner, if present) a bit. And finally, to keep the ring sealing, the top and bottom edges of the piston ring groove were chamfered at 45 degrees in the region near the divider, so that any hammering that did take place could not trap the ring.

H2R engine castings
Three-cylinder Kawasaki H2R engine castings await assembly in 1972 at the factory race shop in Akashi, Japan.Cycle World Archives

Meanwhile Kawasaki, latecomers to the intense GP racing of the 1960s, went in two directions. In the production twins with a rotary intake valve, engineers continued the three-transfer-port orthodoxy of the 1960s, including the porting of the 1967 250cc A1R parallel-twin production racer and, beginning in 1975, the factory-only KR250 disc-valve tandem twin. But in 1970 with its new three-cylinder 500cc H1 and H1R production racer, Kawasaki provided four transfers per cylinder but with the smaller “B” pair fed from the same smallish crankcase ducts as fed the “As.” Both transfer pairs had close-to-flat roofs with little upflow and were aimed back at the non-exhaust cylinder wall.

With the exception of Kawasaki’s three-transfer engines, then, a new orthodoxy had emerged: two pairs of transfers, the “A” pair aimed back some 35 degrees, the “B” pair much less so, and with flat-to-moderate-upflow roofs.

My probe was just a small brass tube with 2mm bore, having a right-angle bend at one end. The other end was connected by flexible plastic tube to the pressure gauge.

Packaging was a big issue because almost all of these engines were production-based. With a clutch sticking out on one side and an ignition and alternator on the other, width was a problem. Kawasaki’s rotary-valve production bikes also had carburetors sticking out to the right and left.

The common packaging solution was to cram the cylinders as close together as possible, even if that meant leaving less room for the ducts connecting the crankcases with the transfer ports. The result was often “elevator” transfers, ducts that went straight up the outside of the cylinder liner, then made an abrupt almost 90-degree turn to enter the cylinder. I used an impact probe and a sensitive pressure gauge to map the flow emerging into cylinders from this kind of transfer. My probe was just a small brass tube with 2mm bore, having a right-angle bend at one end. The other end was connected by flexible plastic tube to the pressure gauge. With a piston held at bottom center by a lump of modeling clay and the headless cylinder sitting on an upper crankcase into which a blower delivered air, I could find out where, over the face of the transfer windows, air was coming out—the regions of positive ram pressure. To my considerable interest, I found that roughly the lower one-third of the window of such an elevator-style transfer duct was not flowing any air at all.

This is analogous to the short-side problem in four-stroke intake-port flow. As the port approaches the valve seat at an angle (which it must do because of the valve springs above), the momentum of the flow carries most of it to the far side of the valve, where it emerges at high speed. Much less of the flow is able to make the sharper turn to flow out of the nearer, or short side, of the valve seat. Coaxing more flow to the short side has been an enduring task of four-stroke cylinder-head porting.

So it was with those straight-up-then-turn-90-degrees transfer ports of the 1970s: Most of the flow, unable to make the tight turn, came out at the top two-thirds of the port, leaving the lower one-third to flow nothing. In return for the narrower engines that resulted from locating cylinders close to each other, designers paid a price—a built-in two-thirds throttle—in the form of those sharp-turn transfer ducts.

Most of the flow, unable to make the tight turn, came out at the top two-thirds of the port, leaving the lower one-third to flow nothing.

Another chronic problem was the need to fasten the cylinders to the crankcase. For years, the usual method was by long studs threaded into the crankcase that came up through the cylinder and head. Where there were studs, there could not be transfer ports.

Because they had just one cylinder, motocross engines were the special case that allowed an easy solution to hold the cylinder in place by a thick, wide base flange, big enough that the short studs retaining it were not in the way of wider and more gently curved transfer ducts. This was especially valuable on reed-valve engines, which needed a lot of width and volume for their intake arrangements.

Suzuki showed a better way to package two-stroke cylinders in its 1972 GT750, also known as the “Water Buffalo.” As each cylinder had just two gently curving transfer ducts, the cylinders were rotated such that those ducts were staggered rather than elbow to elbow. Because that engine was water-cooled, it was simplest to implement all three cylinders in a single casting. Kawasaki would later employ this system on its race-only 1975 KR750.

In 750cc roadracing, Suzuki, by not having to adopt the elevator style of flow-restrictive transfer ducts, was able to make at least as much power at 7,600 rpm from its two-transfer 750 as Kawasaki could produce at 9,000 to 9,500 rpm from the four transfers per cylinder of its 1972–’74 H2R.

In 1972, Kawasaki showed something new in the form of its 750cc H2 triple. Clearly someone had asked the question, “Why should the ‘B’ transfers be only one-third as wide as the ‘As’?” and then had gone on to study the question experimentally. It found improved scavenging with four transfers that were all the same width. In 1972 and ’73, I would learn a lot by trying to understand why.