A Small part of E.T. Westbury's article circa 1951
In practice, the best results are obtained by using a counterweight capable of cancelling out only a portion of the reciprocating weight. The exact amount is often a subject of fierce dispute, but in actual fact it depends on a number of (sometimes incalculable) factors, such as the way the engine is mounted, the moments of inertia in the fixed and moving masses (which influence "critical speeds"), and so on. What really happens is that some of the forces which tend to cause vibration are diverted into other planes, where they may be more tolerable, or more readily absorbed in the structure; in no case is the vibration in a single plane so violent as it would be in an entirely unbalanced engine.
As a general rule, it may be said that engines which are required to run at widely varying speeds require a greater portion of reciprocating weight to be balanced out than those which can be kept running at well above critical speed. In some small high-speed engines it is possible to "get away with murder", by using very sketchy balance weights, or even none at all. This is because the reciprocating parts of these engines can be made extremely light, and their structure resilient enough to absorb vibration; but it should be remembered that the unbalanced forces are still there, and are registered in the mechanical stresses and bearing loads, also that these forces have to be generated by the engine itself, thereby detracting from the power available for useful work.
The complete article is posted on hewmow, it explains how to balance the assembly.
Model A Fords were not balanced, they used a 80 lb flywheel to "absorb" the unbalance, it also only had 26 hp at the flywheel and will break the flange off the flywheel side above 5K