Remarks on clinical "norms".

The " Normal " (Gaussian) Curve Neither the following remarks, nor a textbook discussion (1), nor the efforts of various other writers will quickly disentangle us from the ambiguity of " normal " as used in medicine (and even more vaguely in everyday speech) and " normal " as used by statisticians to indicate the gaussian curve. " Normal " in the statistician's sense means " standard " (norma in Latin = a carpenter's square). The curve is indeed a very useful tool in the manufacturing of various other statistical tools; but this does not prove that it is a picture of any part of the real world, either living or inanimate. It was invented by Gauss (and by several other mathematicians using different methods) to represent the variation between different measurements of the same physical object, measured by the same instrument and technique-hence the term " error curve. " It is well, therefore, to recall a comment on the error curve expressed in different ways by different specialists in the fields of mathematics and physical measurement: " Everybody believes in From the Medical Statistics Unit and Rheumatic Diseases Study Group, Some deletions and editorial changes have been made, with the author's permission, to make the article better fit the present audience. It seemed useful to present this second discussion of the problem of normal ranges, from a different viewpoint, to accompany the paper of Reed and Henry (this issue), especially since the " Notes " are not widely circulated, and since this subject has received much attention in CLINIcAL CHEMISTRY lately. A second purpose is to introduce Dr. Mainland's writings to those of our readers who may not be acquainted with them. He is, as you will see, a gifted expository writer, who ably brings the sometimes esoteric mathematics of statistics into what he calls the " real world "-in the present case, the clinical laboratory. the law of errors, the experimenters because they think it can be proved by mathematics, and the mathematicians because they believe it has been established by observation " (2). When large collections of physical measurements have been carefully examined, divergences from the gaussian curve have commonly been found (2), and there is still less reason to expect that biological (including medical and psychological) measurements would conform to it. Such measurements often form a bell-shaped frequency curve, but bell-shaped is not synonymous with gaussian. …