A brief discussion of physics.
Whenever you measure torque, you must to take into account the distance from the center to the point of measurement in your calculations, because by definition, torque is a measurement of force over distance. If, by whatever means you derive your measurement, you determine a figure of (hypothetically) 100 ounces, you must multiply by the distance from the center to determine the actual torque.
For example, if you used a spring scale, connected to the end of your servo arm to measure the torque (admittedly a fairly crude method, but used for the sake of the example), and the distance from the center of the servo arm to the point of connection was 3/4", then you need to multiply your reading (100 ounces) by .75, to arrive at a torque reading of 75 ounce-inches (or oz/in, sometimes noted as in/oz). If the scale you use is graduated in grams, then your distance measurement has to be converted into centimeters (multiply inches by 2.54) to maintain consistency of units. Further, if your scale is graduated in pounds, you need to convert your reading to ounces (multiply by 16), or change your distance to feet (divide by 12), but not both. Expensive laboratory scales might be graduated in "newtons" which would require you to convert your distance measurement from inches to meters (divide inches by 39.37) All of these will give you a reading of your torque. If you have trouble getting an accurate reading with the scale you have, you can use a longer "arm" to amplify the reading. This would be particularly helpful if you were trying to measure the torque with a scale graduated in pounds. A one foot long arm, securely fastened to your servo arm could be used to give you a torque reading in pound-feet which could then be converted into the more useful ounce-inches (multiply pounds by 16 and feet by 12). Always be careful to measure from the center of the servo hub to the point of attachment, Also, keep your scale (again using the spring scale example) at a right angle to the arm. At best, you can arrive at a fair approximation of the servo torque at that moment. Remember, battery condition, the quality of your connections, friction, interference, and several other factors will influence the "actual" torque at any given moment.