Ultimately, I think the discrepancies you are seeing are due to differences in conditions. As you probably know, the conditions can change the coefficient of friction enormously. Polishing the metal will usually lower the friction, a lubricant will lower the friction, and the temperature will change the viscosity of the lubricant. I'm sure there are others.
My go-to reference for these sorts of things is the CRC Handbook. I have a copy of the 1965-66 edition in my office, and a copy of the 2005-06 edition at home. Unfortunately, the table of coefficients of friction does not have citations to the sources. To take your example of steel on steel, the older edition gives 0.58 for "clean" steel, and about 0.1 for steel that's been lubricated with any of several vegetable oils. The newer edition gives 0.78 for "dry" hard steel. And it can also be anywhere in between if the steel has been lubricated with (for example) benzene, an alcohol, an alkane, etc.
The CRC's are nice because they give some relevant details of the conditions (in this case, lubricant and temperature), and they usually give a citation to original literature with more details about the conditions; I'm a bit disappointed that they don't have the citations for this data. My experience is that the newer editions are usually more comprehensive than the older editions, but that isn't really the case here: these two editions have some overlap with each other, and each has some pairs of materials the other doesn't.
There might be a more specialized handbook that is better for this than the CRC, but this isn't my area, so I don't know what the handbooks would be. I should also point out that the set of materials called "polymers" includes both the low-friction PTFE (Teflon) and several adhesives, so you want to be careful with that.