The only limitation is that sampling frequency must be higher 10 Hz more than two times otherwise you will see a parasite periodicity in the friction coefficient plot.

To my knowlege, a frequency of 10 Hz is not too high in the reciprocating sliding. The system can read the  corresponding values if the sampling rate is higher. The disadvantages of high-speed tribology may be the speed fluctuation and diversity of dynamic COFs.

I agree that 10 Hz may be too high but more depends on the sliding speed level. For a majority of tribological tests this speed must be below 1 m/s, optimum  at  0.5 m/s. Higher speeds are applied usually for hydrodynamic lubrication modes.

Reciprocating sliding s not the best testing scheme but the same could be related to pin-on-disk one since both these schemes are mostly used with tribometers. 

In reciprocating wear test, we can able to run the experiment beyond 1m/s, in my experience. Even at a speed of 0.8m/s the entire system works with some irregularity in the running, but we could able to conduct and complete the experiments at 0.8m/s upto 30 minitutes without any failure.

WHILE DETERMINING THE SPEED OF RECIPROCATION, IT IS IMPORTANT TO GIVE IMPORTANCE TO VIBRATION EFFECTS, WHICH CAN COMPLETELY UPSET THE CONSISTENCY OF RESULTS. ALSO, THIS BECOMES MORE IMPORTANT WHEN TESTING THIN FILM COATINGS LIKE PVD, DLC ETC WHERE THE COATING THICKNESS MAY BE IN THE RANGE OF 3 - 20 MICRONS.

If the test can be run at high speeds then absolutely no disadvantages.

It is impossible to maintain constant sliding speed during reciprocating friction. Each reciprocating cycle should have "stop-start" period with another COF comparing with COF during steady period of that cycle. In some cases it may be big disadvantage.

I follow the opinion mentioned above. The expressed dynamics of high-speed reciprocating wear test (inertia of tribosystem, speed variation and caused by these factors non-stable friction conditions) makes some difficulties of experiments.

There are some conceptual issues to be decided regarding COF in reciprocating tests. What do you intend to obtain: Peak COF from the velocity inversion points (people who look for scuffing use it) or average COF during the whole cycle(s)? The latter may be more important when comparing energy expenditure.