In the experimental study, twelve male subjects held 1, 5, and 10 kg boxes at 0, 20, 40, and 60 cm from the floor in forward-bending and squatting postures. The inclinometers were fixed on the back at L4, Th 12, and Th5.The relation between trunk inclination and the inclinometer reading in measurement data showed concave, straight, and convex curves at L4, Th12, and Th5, respectively, as the theoretical model analysis had indicated. At L4 and Th5, however, the relation could be regarded as linear because the coefficients of determinations (R2) of the theoretical models calculated by the regression analyses agreed closely to those of the simple linear models.The difference between the forwardbending and squatting postures was large at L4, but small at Th12 and Th5.The effect of the weight handled was small and negligible in making these estimates. For ergonomic evaluation of working conditions, it is important to have a clear idea of working postures at actual work places'-4). Observational and goniometric methods are often used to record working postures3-6). Inclinometers are also used for simplified quantitative measurement-10), but the location of an inclinometer on the body is problematic, usually depending on the purpose of the observation-9).To evaluate low back load, the inclinometer is usually fixed on the upper body of the subject to record the trunk inclination (angle between vertical and a line between the acromion and greater trochanter) and a special calibration curve must be developed because the inclinometer does not measure the trunk inclination directly. If the trunk were a rigid component, the estimation would be simple and easy because trunk inclination and the inclinometer reading would show a linear relation, but actual trunk inclination involves trunk bending and pelvic rotation", 12). Most of the trunk bending occurs in the lower portion of the spinal column13, 14), and pelvic rotation is reduced according to the trunk inclination"). The actual relation between trunk inclination and the inclinometer reading should therefore be nonlinear. To what extent and under what conditions can the inclinometer calibration method be linear? This point has been insufficiently studied.In this paper we first try to clarify the relation between trunk inclination and the inclinometer reading by using a theoretical model. We then present experimental data on the effects of inclinometer location, working posture, and the weight handled on this relation.
Theoretical StudyAccording to a statistical analysis by Anderson et aL, pelvic rotation can be estimated by the following equation 15):