The Accuracy of Linear Body Measurements of Dairy Cattle

Abstract

Five measurements[long dash]Wither height, chest depth, body length, and circumference at chest and at paunch[long dash]taken regularly on Holstein cattle in the Iowa State College herd over a period of 16 yrs., were studied to see how much of the variation between measurements made at the same age came from real differences between the animals, how much from the order of the measurement, and how much was random error. Each animal was measured 3 times at each of 7 ages: 6 months, 1, 2, 3, 4, 5, and 7 yrs. The max. number of animals was 367 at 6 mos., there were still 108 at 5 yrs., and 38 were measured at 7 yrs. The variance components for order, for differences between animals, and for error are given for each of the 5 measurements at each age. Avg. differences between the 1st, 2d and 3d measurements at any age were only on the borderline of statistical significance but, in any event, were much too small to have practical significance. The variance from real differences between animals was much larger than the error component at all ages, ranging from between 5 and 9 times as large for body length to between 50 and 100 times as large in the case of paunch girth. Because the error of measurement is small, relative to differences between the animals of the same age, accuracy is increased only a tiny bit by taking more than one measurement at a given age and using the avg. In absolute values the errors were largest for body length, next for paunch girth and heart girth, next for wither height, and least for chest depth. Age had little effect on the accuracy of the measurements, the errors being a bit smaller in absolute figures at 6 mos. than at later ages but larger (insignificantly so) at that age when expressed relative to the amt. of variation between animals. The general conclusion is that errors of measurement are rather unimportant with these 5 measurements. Measuring the animals twice at a given age is worth while only to prevent gross errors, such as misreading the measuring instrument by 10 cm., from going undetected.