Holding Torque

I do not know your application, more information would help. My initial inclination would be to use splines with a thermal fit.

There is some ambiguity in your use of the term "holding torque". If one assumes you are attempting to prevent the rotation of an attachment with respect to the shaft on which it is mounted, then the resistance to turning will be the resistance to shearing encountered at the interface plane between the attachment and the shaft. Resistance to shear is going to be a function of the cross-sectional area of whatever scheme you are evaluating.

Theoretically, it would be possible to provide the same shear area with a knurl as one would achieve with a square key, but, due to the way knurls are usually cut into the surface of the shaft, there is a good chance that it would be stronger in one direction than the other. Practically, it would be very difficult to insure that one had achieved sufficient shear area with a knurl to equal what ones achieves with typical square key design practices.

Case in point- I can often twist a knob off a knurled shaft by hand, and I am no superman (granted, we are usually talking a plastic knob on a metallic shaft in this sort of case). I have never been able to twist a keyed attachment off a shaft by hand. This does not necessarily mean that the keyed attachment is stronger- just that it has been designed to provide higher resistance to shear...

My question was more general in nature and not tied to any specific assembly or size. By holding torque I did mean retention of an object on a shaft, such as a gear and shaft assembly.

I had wondered if a knurled shaft design was better than a keyed shaft design for any particular size group or if either design was always considered better.

Understand that it basically boils down to the amount of shear force but didn't know if shearing one larger surface area, such as a key, was more difficult than to shear multiple, shallower, smaller surface areas like a knurl.

Also wanted to learn if both features were implemented in a design what the total contribution would be. For example, if the knurl design calculated out to be capable of 500 lb-in retention and the key's capability was 650 lb-in retention would total retention then be 1150 lb-in or something less?

Wondered if this could be calculated as opposed to testing. Reason for this is that I am dealing with many different sizes.

Appreciate any comments.