The usual method is to generate torque and RPM points using a number of different "props".
The props usually have no pitch so that there is no airflow to disturb the measurements. Basically paint stir sticks.
Motor is mounted on an axle and the torque determines how much it twists.

Some pictures here:

https://www.rcgroups.com/forums/showpost.php?p=35713764&postcount=669

Lots of details I am sure in getting it all to work properly, but that is the general principle.


There are also picture there of other methods,  for example using magnetic loads that you can vary and a load cell to directly measure the torque.

I have thrust finder machines  that  are  very good at this, as comparing  engines to engines

Randy

Thanks for the clarification, Pat.
I'm trying to understand about prop selection and this is one of those areas in which more info will probably cause my question to change, but here goes:
Case in point: a Golden Bee for some stuntability, for which a Cox 5x3 prop has been most often recommended, I must assume from experience.
Mine (twin bypass, standard head, 1 gasket) will turn a 5x4 or a 5.5x4 prop on 15% nitro to ~17K static (2-300 more on 25%), and Waring, in his report, says peak power is at 13-15K.
Which means those props peaked would be well past max power in the air (I think Chinn said 1-1.5K unloaded increase). Of course the engine can be richened to bring launch rpm down to 12.5-13K to keep unloaded rpm near peak hp.
Why not choose a larger prop that would unload to the sweet spot but still has "breathing room" in the air?
Seems it would give better mileage, not being richened so much, if nothing else.

Model Engine News has an article about the different methods sometimes used to measure HP at http://modelenginenews.org/techniques/testing.html and the simplest is the last one on that page by using calibrated props. For general use at home I'd think that a range of closely similar props would give good comparative results for a large range of engines although not quite as accurate as calibrated props. Many years ago I used those graphs to make up a spreadsheet so that, by entering the revs achieved on any one of those props, it would show the HP and torque developed. 

To Randy's point I would have to disagree a bit. Thrust is as much about the prop as the motor.
But if you have "known" props then you can compare motors just by comparing RPM, no need to measure thrust.
But if you want an actual HP curve you need multiple data points, so you need multiple loads and some way to measure the torque absorbed.

Disagree  all you want, but your are wrong ,thrust is not as much about the prop as the motor, "if" you have a more powerful motor turning the exact same prop you will more than likely get more thrust which equals more  power, its all about the  motor, not  the prop as far as this particular test goes
The MOTOR makes the thrust by having power to turn the prop, the single prop does not make more or less thrust, its the motor that does that, as long  as you use apples to apples compare
I personally have a set of  known props  that I use  to  plot a torque and HP curve, but  the  thrust finder is, as I said a useful tool.

http://bj-model-engines.com/AmericanHobby/Dyna.html

The company that makes the device I sell  also makes a  dyna-meter  whick measures torque


Regards
Randy

I did say that you could use a thrust stand if, like you, you have "known" props. But RPM would tell you the same thing.

But if you don't have any known props and the goal is to measure the HP curve then the dyno is the tool you need.
Or you can use the curves and the RPM.

Like this one, Steve?

Should be fairly easy to compare the wattage drawn by an electric motor at a specific rpm for half a dozen or so blank props, like the Kavan test props, if you want to make your own.  It's easier to work with using blank props because regular props make a heck of a breeze.

Years ago a friend, Glenn Dye, built several TWA rear exhaust combat motors for me.  We tested several of them using Kavan test props he had.  They had a very interesting torque curve- it started low, of course, a smooth curve up to about 1.2 hp @18,000 rpm and then turned into a very flat curve that topped out at 23,000 and then dropped back to 1.2hp @ 25,000.  Different props had little effect on the speed, so I went with the ones that held speed in maneuvers the best- an 8.5/6 pylon prop if I remember right.  3 minutes of high excitement!