This section is from the book "Cyclopedia Of Architecture, Carpentry, And Building", by James C. et al. Also available from Amazon: Cyclopedia Of Architecture, Carpentry And Building.
In drawings which are not made on the slate the following method of testing proportions is usual. With the arm stretched forward to its greatest length, hold the pencil upright so that its unsharpened end is at the top. Move it until this end coincides with the uppermost point of the object. Holding it fixed and resting the thumb against the pencil, move the thumb up and down until the thumb nail marks the lowest point of the object. The distance measured off on the pencil represents the upright dimension. Holding the pencil at exactly the same distance from the eye, turn it until it is horizontal and the end of the pencil covers the extreme left point of the object. Should the height and width be equal, the thumb nail would cover the extreme right edge of the object. If the width is greater than the height, use the height as a unit of measurement and discover the number of times it is contained in the width. Always use the shorter dimension as the unit of measurement. The accuracy of the test demands that the pencil should be at exactly the same distance from the eye while comparing the width and height. In order to insure this, the arm must not be bent at the elbow and must be stretched as far as possible without turning the body, which must not move during the operation. The distance from the eye to the object must not change during the test, and the position of the eye and body is first fixed by leaning the shoulders firmly against the back of the chair and keeping them in that position while the test is taking place. It is equally important in both the upright and horizontal measurement that the pencil be held exactly at right angles to the direction in which the object is seen; i.e., at right angles to an imaginary line from the eye to the center of the object. In either position the two ends of the pencil will be equally distant from the eye. The test should be made several times in order to insure accuracy, as there is sure to be some slight variation in the distances each time. Avoid taking measurements of minor dimensions, as the shorter the distances measured the more inaccurate the test becomes. At the best measurements obtained in this way are only approximately correct, and too much care cannot be taken in order to render the test of use. Applied carelessly, the test is not only valueless, but thoroughly misleading. When there is any great conflict between the appearance of the object and the drawing after it has been corrected by the test, it is often safe to assume some mistake in applying the test and to trust the eye. In such a case the test may be tested by the use of the slate. A few lines and points will be sufficient to indicate the width and height on the slate, and the relative proportions can then be calculated.
The plumb-line affords another method of testing. A thread or a string with any small object for a weight attached to one end, is sufficient. Hold the string so that it hangs vertical and motionless, and at the same time covers some important point in the object. By looking up and down the line the points directly over and under the given point can be determined and the relative distances of other important points to the right and left can be calculated. The plumb-line will also determine all the vertical lines in the object and help to determine divergence of lines from the vertical.
A ruler, a long rod, or pencil held in a perfectly horizontal position is also of assistance in determining the width of angles and divergences of lines from the horizontal.