were estimated from a knowledge of only the parameters of stature in the survey concerned. (This is a more rigorous test than the 5th and 95th %iles commonly used in design work.) The errors that accrued were random and conformed approximately to a normal distribution with parameters of –3 [13] mm. Ninety-three per cent of the errors fell within the range of ±25 mm. In many cases the estimates were within the confidence limits of the original survey.
2.4.2 Accuracy
What accuracy is actually required in anthropometric data? This is a very difficult question which must be studied at several different levels. In the purely formalized statistical sense we may consider what percentiles a given percentile that is erroneously quoted actually represent, e.g. if a 95th %ile was in error, the figure quoted might in truth represent the 93rd or 98th %ile, with a consequence of mismatching a greater or lesser percentage of the target population in the design. In the validation study the estimates of the 1st and 99th %iles were checked in this way—on average the estimates would have included 96% of the population as against 98% for perfect data. It is more informative, however, to consider the likely errors of prediction alongside those that might arise in other ways. The human body has very few sharp edges—its contours are rounded and it is generally squashy and unstable.
The consequent difficulty in identifying landmarks and controlling posture makes it virtually impossible to achieve an accuracy of better than 5 mm in most anthropometric measures—and for some dimensions the errors may be much worse (sitting elbow height is a notorious example). These errors, however, pale into insignificance in comparison with those that might occur in the application of even the most accurate tables.
In applying anthropometric data we commonly need to make corrections for clothing, postural variation, and so on (see below). These corrections, although they are better than arbitrary, will usually be inexact—as (perhaps more importantly) will be the anthropometric criteria we apply to define a match. Take the case of seat height which we discussed above. The sensations of discomfort which the user experiences will become progressively more pronounced as the height of the seat exceeds his or her popliteal height. But there is no obvious and clearly defined cut-off point at which we should say ‘thus far and no further’. In practice, anthropometric criteria are almost always ‘fuzzy’ in this way.
There are doubtless certain safety-critical applications in which accuracy would be at a premium. But experience indicates that these are the exception rather than the rule. In practice there would be few everyday problems requiring an ergonomic specification to an accuracy of more than 25 mm. We could call this the anthropometric inch.
2.4.3 Clothing corrections
Most anthropometric measurements are made on unclothed people; most products and environments are used by clothed people. The data tabulated below are for unclothed people. So before applying these data to any particular problem it will in general be necessary to add an appropriate correction for clothing. (It makes sense to do it this
way, rather than to quote figures with clothing corrections already added, because the magnitude of the correction may vary greatly depending upon the circumstances.)
The most important of these corrections is an increment for the heels of shoes which must be added to all vertical dimensions which are measured from the floor.
The thinnest pair of carpet slippers has a heel height of only 10 mm. The most outrageous pair of high heels may add 150 mm to a woman’s height. A typical heel height for men’s ordinary everyday shoes and for women’s flats is around 25 mm±5.
Women’s shoes (and men’s shoes to a lesser extent) are subject to periodic changes in fashion. The products and spaces we design will presumably remain in use over several of these fashion cycles. In theory therefore we should base our heel height correction on the midpoint about which these cycles oscillate. In theory also we should add an increment to the standard deviation of our dimension as well as the mean, to allow for variability in heel height. In practice, however, variability in heel height is small compared with anthropometric variability and a uniform increment to all percentiles will be adequate.
Taking one consideration with another, the following corrections would seem appropriate for shoes worn in public places on formal and semi-formal occasions:
§ for men, add 25 mm to all dimensions;
§ for women, add 45 mm to all dimensions.
Corrections for situations where other types of footwear are the norm should be made on an ad hoc basis. Other clothing corrections are in general likely to be small—
except for very heavy outdoor clothing or for specialized protective gear, etc. Some examples are given below when discussing individual body dimensions.
2.4.4 Standard anthropometric postures
Most of the measurements described below (and likewise, those given in Chapter 10) were made in one of two standard postures.
In the standard standing posture the subject stands erect, pulling himself up to his full height and looking straight ahead, with his shoulders relaxed and his arms hanging loosely by his sides. He stands free of walls, measuring instruments, etc.
In the standard sitting posture the subject sits erect on a horizontal, flat surface, pulled up to his full height and looking straight ahead. The shoulders are relaxed, with the upper arms hanging freely by the sides and the forearms horizontal (i.e. the elbows are flexed to a right angle). The height of the seat is adjusted (or blocks are placed under the feet) until the thighs are horizontal and the lower legs are vertical (i.e. the knees are flexed to a right angle).
Measurements are made perpendicular to two reference planes. The horizontal reference plane is that of the seat surface. The vertical reference plane is a real or imaginary plane which touches the back of the uncompressed buttocks and shoulder blades of the subject. The seat reference point (SRP) lies at the point of intersection of these two planes and the median plane of the body (i.e. the plane that divides it equally into its right and left halves).
People rarely use these upright positions in everyday life. In practice this may not be so much of a problem as it seems, since we shall commonly set our criteria in such a way as to take this into account. There are circumstances, however, where it may be
appropriate to make a nominal correction for normal sitting slump. Where this is the case, as a rough approximation for adult populations, subtract 40 mm from all percentiles of relevant sitting dimensions.
2.4.5 Defining the target population
The principal factors to take into account when defining a target population of users, for the purpose of selecting an appropriate source of anthropometric data, will in general be: sex, age, nationality (or ethnicity) and occupation (or social class), generally in that order of importance. Where the target population includes children, then age will take first place. The presence of ethnic minorities in a population sample tends to be more of a problem in theory than in practice. As a general rule of thumb, percentile values are unlikely to be affected to any significant extent until the minority group reach 30% of the total or more. Again, however, there may be exceptions for certain safety-critical applications (e.g. guarding of machinery; see Thompson and Booth 1982).
At the end of this chapter you will find a table of best estimate figures for the bodily dimensions of the adult population of the UK aged 19–65 years (Table 2.3). In the chapters that follow we shall treat this as the standard reference population on which we shall base our design recommendations and other anthropometric calculations. Data for other target populations and details of sources, etc., will be found in Chapter 10.
2.5 An annotated list of body dimensions