An optician measures the curvature of a lens surface with a device called a lens meter. A physicist makes the measurement with a spherometer, of which the lens meter is a development. A modification of either instrument may be used for measuring the curvature of a bow limb. The device may be called a curvometer or bend meter.

For a number of years I had been using a simplified spherometer in gaging the curvature of a bow limb. Since the back of a bent bow is essentially the section of a cylinder, the device requires only three contact points; the two outer ones are fixed, and midway between them there is an adjustable one, which is nothing more than the tip of a micrometer screw. In use, the fixed points are placed on the back (or belly) of the bow, and the micrometer screw is adjusted until its point makes contact at the location where curvature is being measured. From the reading of the micrometer and the distance between fixed points, the curvature can be computed.

The modified spherometer just described—which, incidentally, might be called a cylindrometer—will not be discussed further, since it was superseded about two years ago by the present bend meter. In it, the micrometer screw is replaced by the well-known dial gage, generally in use by machinists. The gage has a "feeler," the displacement of which, in the usual form, is shown in thousands of an inch on a graduated scale. Certain experiments which I had in mind indicated the need for a more sensitive gage, so I decided on the type which shows ten-thousandths of an inch per scale division, each about an eighth of an inch long. Figure 1 shows the indicator with its somewhat crude mounting, together with a short section of a composite bow. The dial is 2½" in diameter, and the distance between the fixed points is 1.62". It measures very small changes in curvature, as well as very small curvatures.

My usual procedure in tillering a bow is first to adjust the limbs by visual inspection, never bending them beyond their braced condition. Then I mark ten or a dozen points along the back of each limb, and obtain the curvometer readings at all of them. Having unbraced the bow, I repeat the readings, and obtain the difference corresponding to each of the points. These differences should be constant for a bow with uniform curvature, i.e., with limbs that bend in circular arcs. If they are not, scraping at the proper places soon makes them so. If parabolic, hyperbolic or even elliptical bending is desired, tables of differences must be prepared as a guide in tillering. Circular bending has proved itself satisfactory through the years, and tillering to constant differences is obviously as simple as the process can be made.

It should be noted that up to this point the bow has not been drawn, only braced. We are now ready to set it up at half draw, and repeat the measurements for constant difference. In general, there will be no need for retillering, except for adjustment to a desired weight. Such procedure assures against overstraining any weak part of a limb unless the inequalities at the start are greater than any competent craftsman would tolerate on visual inspection.