The ergonomic study of engineering drawings

Applied Ergonomics 1971,2.3, 162-170

The ergonomic study of engineering drawings

J. Spencer and N.D. Milnes-Walker.

Department of Psychology. University of Bristol.

This article describes an investigation of engineering-drawing comprehension. A variety of subjects was used so that results would have a wider application.

Productivity in manufacturing is partly due to the gener- ation of suitable production instructions. Probably the most common form is the engineering drawing. In spite of their critical importance to ensuing manufacture the needs and the effectiveness of engineering drawings have received little systematic study. Codes of practice and standards exist with- in firms and at national level, for example BS 308, (1964), but few formal enquiries appear to have been made to establish the suitability of the chosen standards.

One reason for this lack of investigation is that the cost accounting typical of most firms does not accurately allocate scrap production costs to the drawing office, so that the real cost of badly comprehended drawings is hidden. Turner (1967), for example, referred to a survey of errors on drawings which suggested that as much as 70% of scrap production in the engineering industries is attributable to drawing production errors or drawing comprehension errors.

Secondly, in mass production industries, any errors, ambiguities or omissions in drawings of a new pr6duct are discovered during prototype production, the cost of which is fairly small compared to the costs of the main production. Against this it may be argued that the ability to produce efficiently and rapidly on a small batch basis is vital to a country which depends on exporting technically advanced, and therefore frequently changing, products. If this argument is accepted, it is clear that-the comprehensi- bility of engineering drawings is crucially important to production success, and warrants investigation.

The characteristics of an engineering drawing

What must an engineering drawing as a production instruction tell its reader? Consideration suggests that it must inform about three basic aspects of the object drawn;

1. Shape or form - a question of pure geometry. 2. Sizes and permissible limits on sizes - a question of

geometry, logic and manufacturing process. 3. Production information about specialised features of

the drawn object - surface f'mish, heat treatment, surface preparation for painting or plating, for example.

This information need not be completely given on one drawing. What information and how much of it to include

will depend on the recipient of the drawing: designer, tool- maker, machine shop charge hand, salesman, or customer. Unfortunately the possible users of a drawing are very varied and to meet their differing individule requirements satisfactorily means that considerable duplication of drawing may occur. This in turn means careful control of drawings issued to different users to ensure that later design modifications are added to every relevant copy of the drawing. Because of the administrative difficulties and costs which result from this solution, the/e is a strong demand for the production of a single comprehensive drawing containing all the information in a form comprehensible to the majority of potential users.

But a single comprehensive drawing is obviously very difficult to produce for at least two reasons. Firstly, the use of colour is not acceptable in most cases, because the information is lost in monochromatic copies. Booker (1963) mentions the universal use of colouring to aid comprehension on engineering drawings during the 19th century prior to the advent of rapid copying facilities. This seems yet another case therefore, where a technological advance represents improved economy and convenience in a very limited sense only. Secondly, the necessary precision of a proportion of engineering components has increased over the last four or five decades, so that a more extensive spec- ification has to be presented on the drawing. This further increases the information density on the drawing.

In spite of the difficulties, it is obviously useful to explore ways in which the comprehensible drawing might be achieved. Correct presentation of geometry for the user rnight so simplify the layout that additional production and dimension information can be included, without adding to the readers' confusion. Attention to the logic and presentation of dimension information might considerably aid comprehension without hindering the ready appreciation of the geometry of the drawn object.

Before describing attempts to study some of the problems, it should be emphasised that spectacular improvements are unlikely. The various codes of practice in current use embody the everyday practice of draughtsmen, and so can be said to represent at least an acceptable set of rules empirically evolved over about a century. It is unlikely that re-orientation of attitude to the problems, as a result of

162 Applied Ergonomics September 1971

applying psychological principles of perception or cognition to them, will suggest radically different ways of presenting production instructions.

However, it is very likely that improvements in detail may add up to a substantial decrease in reading time and reduction in scrap produced. Finally, it should be realised that what is difficult for the reader may also be difficult for the draughtsman. His comprehension of the drawing, as it develops, vitally affects the possibility of errors which go undetected until a later stage of production. Turner (1967) refers to a survey showing that drawing offices required 60% more paid hours in 1954 than in 1929 as a result, among other factors, of greater complexity of the drawn object. He also quotes £50 as a typical cost for drawing of an aver- age mechanical item. Clearly quite small improvements in this area could more than justify the research money spent on achieving them.

The investigation of drawing comprehension It was stated earlier that a variety of users read drawings.

Each reader examines a drawing to solve some particular problem or set of problems. Clearly, the actual problems that might be settled by examination of a drawing can vary over an enormous range in terms of content (whether geometrical, dimensional, tolerance or fabrication detail, for example) and in terms of complexity, (whether this car is too long for my garage or will the Mark II gear box fit into the Mark I engine casing, for example).

In view of this variety of possible objectives that people have when they read drawings it is clear that no simple definition of comprehension will be staisfactory. This must be defined in general terms, as the degree of success achieved in solving some problem whose answer is obtainable from the draw!ng. The difficulty when planning an experimental study is in choosing what sensible operational definitions to limit 'some problem' to a small number of fairly specific situations of fairly general occurrence, so that the experimental results will have both validity and applicability.

Whatever operational definition is finally adopted, it must suggest criteria by which degree of comprehension may be assessed in some objective and preferably quantitative form. A definition which commits the experimenter to some form of subjective evaulation of the 'goodness' of the solution to the problem introduces severe methodological problems.

Assuming that acceptable operational definitions have been agreed and that degree of corn prehension can be measured satisfactorily, two further major problems must be solved before experimental details can be considered.

First, what sort of drawing users should be considered suitable as experimental subjects? Ideally, a variety of subjects will be preferable because the results will have more generality, and more information will be gained by compara- tive analysis about the role of specialised types of experience in determining the level of comprehension by users. As a special case, the completely naive individual must be considered. It might be argued that to expect a completely untrained person to make sense of a drawiiag is unrealistic. However, it is not unrealistic to expect some aspects of a drawing to be comprehensible without extensive training. Admittedly, a drawing contains information coded in specialised ways, but a lot may be learned about the efficacy of the codes used by studying the n~ive response

to various parts of the codes, especially if alternative codes are investigated.

Secondly, although it was stated earlier that engineering drawings are ideally produced in accordance with various standards and codes of practice, there is no universal agreement about all aspects of drawing presentation.

In BS 1100: Pt. 9:1944 the following paragraph appears on p. 16:- 'Many of these practices and conventions have been in more or less common use for many years. There are certain controversial practices upon which agreement has been difficult due to long established usage in individual organisations and industries, where it is thought that radical changes would cause disturbance outweighing the advantages sought by the change. Details which come readily to mind in this respect are methods of projection, decimal versus fractional dimensioning, methods of expressing limit- ing dimensions and machining symbols.' In an investigation reported by Spencer (1963), all four details referred to were treated differently by different designers, which suggests that change to increased standardisation is very slow.

Consequently, alternative ways of presenting the same message exist.This suggests that a fruitful method of investigation would be to study the degree of comprehension achieved by different types of user when each tries to solve the same problem presented in accordance with the differ- entcodes. By adopting this method of investigation it should be possible to examine the effects of experience, inter- individual differences, and presentation techniques on comprehension. The interactions between these three sources of variation should clarify the sort of processes that undefly comprehension and thus perhaps suggest untried presentation techniques which require less experience or training for the same degree of comprehension than existing techniques.

The experimental method used to study comprehension consisted of asking a variety of types of drawing user to solve a series of specialised problems using drawings which vary in the techniques used to present their information. Simple measures of degree of comprehension were used as far as possible.

By its nature, this method of investigation is piecemeal and therefore requires that the problems were investigated in a sequence; first problems concerning the presentation of the geometry of objects, next, problems concerning dimensions and tolerances (which are still under investigation) and the third stage of investigation covered ancillary information. The final stage is planned to verify the results of previous stages when all the beneficial coding is combined into a single drawing. The choice of sequence was chosen because an appreciation of shape is fundamental to an appreciation of dimensional information which in turn is important to the appreciation of production information. This article outlines some experiments and the results obtained in a study of shape comprehension. A later article will describe experiments and results from a study of dimension information presentation.

The experimental study of shape comprehension The comprehension of shape was operationally defined

as the ability to reproduce in model form a shape that was specified by a drawing. It was not intended to measure model making skill, the task performed by subjects had to be simple and easily learnt, yet the shape had to be complicated enough so that in a series of trials for one subject the

Applied Ergonomics September 1971 163

shape was sufficiently different each time to eliminate effects due to memory. The task finally selected required subjects to assemble the specified shape from pegs and rivets. Fig 1 illustrates two of the types of object used in these studies. The rivets signify blanking plates which covered holes in the body. On the drawings nothing would be shown in the corresponding positions. Figs 2a and 2b illustrate the types of drawings which were presented to the subjects.

An assembly task is attractive because it permits time and error scoring and because observation of the subjects' activities during assembly reveals a lot of information, by inference or supplemented by verbal comments of the subjects, about how the subject thought that the drawing was laid out. As a task it is not totally unrealistic but of course it samples only one type of behaviour that occurs in the attempt to comprehend the shape of an object. Alterna- tively, for example, subjects might have been asked to sketch or model in plasticine the object shown on the drawing. This would not be staisfactory because of uncertainty as to whether the sketch or model was correct but badly presented or the drawing wrongly interpreted.

The first series of experiments was concerned with square prismatic shapes. Because this particular shape may have influenced the results obtained, a second series of experiments used both spherical and truncated spherical shapes. This series was intended to verify results obtained earlier on the simpler shape.

The table below shows which subjects were studied with which type of model assembly.

Subjects Grammar Art Draughts- and students men secondary modern boys and girls 13yrs 15yrs

Inspect- Machine ors shop

personnel

Object

Square prism X X X X

Spheres X X X

Schoolchildren and arts undergraduates were chosen to represent the naive users with differing degrees of development and intelligence. Draughtsmen were chosen as repres- entatives of highly experienced users in their capacity as drawing producers. All draughtsmen were employed on mechanical component drawing. The inspectors and machine shop personnel were chosen to represent experienced readers of drawings but with differing purposes in their normal use of drawings.

The procedure adopted with every subject was the same for all experiments. The aim of the experiment was briefly described, after which the subject was given the components needed to make the models. When he was satisfied that he understood what he had to do he was given the drawings, one at a time, and told to model the drawn object as accurately and rapidly as possible. No time limit was imposed and a minimum of additional information was given subjects after they received a drawing. As far as possible all subjects worked in a quiet well4it room free from disturbance Although encouraged to comment verbally on their activities, very few subjects did so until the end of the experiment.

Because the object of these experiments was to throw light on shape comprehension, the drawings used varied only in the method by which the geometry of the object was portrayed. Most of the experiments involved four types of presentation, two being standard orthographic projections, namely First and Third Angle, and two being Representational, namely Oblique and True Perspective. It is important to note that with representational portrayal of the shapes used in these experiments, two views only are necessary to fully describe the shape. Five views were necessary with the orthographic projections.

As a check on a possible interaction between shape comprehension and dimensional information acting as visual noise, half of the drawings of prism shaped objects given to the draughtsmen and university students were fully dimensioned, while the other half had no dimensions. In all other cases the drawings showed no dimensions.

The first series of experiments using university students and draughtsmen to comprehend drawings of prismatic

Fig 1 Prism-shaped and spherical objects used in shape comprehension.


I

I

3.000 dia. l, o,o,,ol

© 0

• 6

T

Detail of 4 drilled holes Detail of S pegs (assembled)

Fig 2a First angle orthographic projection. The centre object viewed from the left is shown on the right side of the drawing. The centre object viewed from the right is shown on the left side of the drawing. The centre object viewed from above is shown at the bottom of the drawing.

P

Fig 2 b Representational drawing. True perspective views of a square prism assembly.


objects has been described in detail by Spencer (1965). The principle quantitative results are shown in Tables 1 and 2 in giving the median assembly times and numbers of successful assemblies expressed as a percentage of the total number of assemblies attempted. The tables contain data for all the experiments carried out in the area of shape comprehension.

Table 1 Assembly perform ance with square prism objects

The most striking feature of the results for square prism assembly is the poor performance when orthographic drawings are used. Measured in both time and proportion of correct assemblies, performance is much poorer compared with performance on representational drawings. A more important result which can be seen by inspection of Fig 3

Subject 1st angle group

Median % assembly Correct time

Projection system

3rd angle Isometric

Median % Median % assembly Correct assembly Correct

assemblies time assemblies time assemblies

True perspective

Median % assembly Correct time assemblies

Secondary

Modern

Grammar

University

Draughtsmen

Age

13 yrs (n=10)

15 yrs (n=10)

13 yrs (n=lO)

15 yrs (n=10)

(n=20)

(n=20)

28'47" 10o/0 18" 15" 0o/0 2'53" 30o/0

24' 16" 0o/0 13'32" 0o/0 2'23" 30o/0

5' 7" 10°/0 4'23" 0% 1'44" 40%

10'41" 0o/0 12'35" 00/0 2' 0" 20o/0

7' 18" 17%* * 4 '35" 36%* * 1 '48" 68% * *

7' 0" 33.3o/0 4'37" 37% 1'37" 85%

2'30" 0°/0

2'31" 30°/0

1 '52" 50°/0

2'34" 7O%

1 '38" 68%* *

1 '40" 85%

**Based on performance with dimensioned and undimensioned drawings. Times and percentages rounded to nearest unit. Each subject saw one drawing on each projection system.

Table 2 Assembly performance with spherical objects

Subject group and

condition

Projection system

1st angle 3rd angle Isometric Section A Section B

Median % age Median % age Median % age Median % age Median % age assembly Correct assembly Correct assembly Correct assembly Correct assembly Correct time assembly time assembly time assembly time assembly time assembly

Draughts- Sphere 4'36" 89% 2'28" 57% 3'29" 21% 4'51" 55% 5'39" 80%

men Trunc- (n=19)

ated Sphere 3'10" 70% 2'29" 95% 6'37" 39% 6'39" 73% 7'27" 39%

Inspect- Sphere 9'26" 50°/0 7'19" 40°/0 4'26" 10°/0 11 '48" 13% 6'34" 37.5%

ors Trunc- (n=10) ated

Sphere 3'34" 70% 5'46" 60% 7'41" 10o/o 8'12" 0% 4'17" 0%

Machine Sphere 7'41" 78% 5'32" 80°/o 3'59" 20% 19' 7" 25% 7'39" 44%

Shop Trunc- Operat- ated

ives Sphere 3'38" 80% 5'39" 70% 6'49" 40% 10'37" 30% 15'59" 22% (n=lO)


2211

1938

1727

162S

1456

1332

1223

I095

1080

900

800

700

t -

O u

v1 600

c

E I-

SO(]

400

300

200

100

o•o trOts age

Projection

Subjects

Ke___Z

I

I

= = = = = ! I

90 I(30 70 I00 I00100 70 70

3 I T I 3 I T Secondary modern

13. yrs I 5yrs

Project ions:- I= First angle

I I I I I I I I

gO ICE) 60 50 100100 80 30

I 3 I T I 3 l T

Grammar

13 yrs 15 yrs

3 = Third angle I = Isometric

X

It i I I I

83 64 32 32

I 3 I T University

I I I I

66 63 IS 15

I 3 I T

Draughtsmen

T = True perspective

Fig 3 Median, upper and lower quartile assembly times and percentage error incidences for square-prism assemblies by six groups of subjects, using four projection systems.


and Table 1 is that the differential effects of age and experience are considerably reduced when representational drawings are used as compared to the use of orthographic drawings. This suggests that the representation projections are a more readily appreciated method of coding the geometry of a solid object than are the orthographic projections. This is not unexpected but it is pleasing to show experimentally what many people have considered axiomatic in the absence of published evidence.

The second feature of the results is that in most cases the time taken is less with Third Angle drawings than with First Angle drawings although the incidence of correct assemblies are not very different. Again, it should be noted that this difference tends to be greater as experience or ability diminishes. This suggests that although in terms of geometry the two projections are equivalent, they are not so in terms of human comprehension. An attempt to illustrate this point more forcibly is provided in Table 3. This shows the percentage proportion of assemblies which were commenced on the basis of a particular interpretation of the lay-out of views on an orthographic drawing. Thus all subjects had been told that the various views on an orthographic drawing were laid out systematically. The draughtsmen naturally knew this and realised that they first had to establish whether the projection was First or Third Angle before commencing assembly. The naive subjects however nearly always commenced assembly quite soon after receiving an orthographic drawing and revealed by their sequence of moves what assumptions they were making about the relationship between the views on the drawing.

Several possible relationships could exist between five views of an object, but only two are of practical importance, namely those which correspond to First and Third Angle lay-outs. Assembly attempts which did not conform to one of these two assumptions have been classified as 'other' in Table 3. In most cases, especially with younger subjects the behaviour classified as 'other' was not consistent. The interesting point about the results given in Table 3 is the high proportion of Third Angle assemblies in spite of no experience or training in the geometry of this projection. This might be described as a stereotype in view of its frequency of occurrence over a wide range of ability and experience. It is interesting to note that Third Angle lay-out is equivalent to an 'exploded development' of an object. In other words people seem to apply every-day experience with solid objects to this problem of presenting

Table 3 Interpretation of the layout of orthographic drawings by naive subjects, expressed as percentages allocated to possible interpretations

Subject Interpretation chosen by subject Total

group 1st angle 3rd angle Other

Secondary 13 yrs 8.3% 58.3% 33.3% 99.9% Modern 15 yrs 5.0% 60.0% 35.0% 100.0°/0

Grammar 13 yrs 5.0% 60.0°/0 35.0% 100.0% 15 yrs 0% 55.0% 4 5.0% 100.0°/0

University 3.8% 92.4% 3.8% 100.0%


a solid object on a two-dimensional surface. Thus the adjacent edges on two views of the same object are treated perceptually as being related or common.

The presence or absence of dimension lines on a drawing had less effect on performance than had been expected. Also the presence of dimensions and their associated dimension lines seemed to be less detrimental to comprehension of shape in representational drawings than in orthographic drawings. However, the differences that did occur were not significant statistically.

Effects of shape complexity

It could be argued that the nature of the square prism assembly task pre-disposed subjects to behave in the ways just described. Thus the orthographic views of a surface with a hole or a projection of the same diameter are identi- cal and to establish whether in fact a hole or a projection has been drawn the reader must cross-refer to a view at right-angles to the one posing the problem. This cross- referencing is unnecessary with representational projections because none of the principal axes of the drawn object lie in the planes of the paper. Therefore one would expect representational projections to be superior for this particular assembly task. More complex shapes would not yield the same results. Hence the second set of experiments was carried out using spherical objects, which may be regarded as the most difficult of compound surfaces to depict in two dimensions.

Examination of Table 2 shows that a higher number of successful assemblies with similar assembly times was achieved in the orthographic than with representational projections, that is the results for spherical assemblies reverse those found for square prism assemblies. This justi- fies the reason for examining performance with objects other than square prisms, but it casts doubt on the generality of the advantage previously claimed for representational projections. Inspection of Fig 2 suggests a reason for the occurrence of the apparent discrepancy between the results of the two sets of experiments. It is immediately clear that with a curvilinear object, only a small part of the total surface area of the object would lie in the plane of the paper on which it is represented. The remainder of the surface and its topography would be represented obliquely, and approx- imates to a sort of compound oblique projection of the surface. Thus the difference between orthographic and representational projections in this situation lies primarily in the number of views presented. The individual views on both types of drawing are basically similar, but the subject is given four views on the orthographic drawings and two on the representational drawings.

Interpreted in this way the so-called orthographic drawings correspond to the representational drawings plus 100% redundancy of information. If this is correct, then on the basis i) that spherical objects are more complex than prismatic and ii) that redundancy will aid comprehension in situations as difficult as those presented to the subjects one would predict the following results:

Performance will be better: l) for prisms than spheres when both are presented

representationally. 2) for spheres than prisms when both are presented

orthogonally. 3) for orthogonally presented spheres than for

representationally presented spheres.

Examination of the bottom row of Table 4, shows that there is some statistical support in favour of 1) and 2), while 3) le is clearly supported for assembly success but the result t is inconclusive for times. The comparison can only be regarded as illustrative rather than conclusive because two different samples of draughtsmen were involved. Although it is unlikely, it is possible that subject differences could have inflated the differences shown.

The effects of providing a reference feature on an otherwise 'meaningless' object was studied in these experiments by comparing performances using spherical assemblies and truncated spherical assemblies. One of the latter is illus- trated in Fig 1. It was expected that truncation, by providing an orienting feature on the otherwise uniform surface of the sphere, would improve performance. The results show that for draughtsmen, inspectors and machine shop operatives, truncation improved performance times when First Angle drawings were used but not when Third Angle drawings were used. The effect of truncation on assembly correctness is in most cases slight and variable.

The results for two experimental methods of presenting a spherical surface are shown in Table 2 under column head- ings Section A and Section B. It can be seen that they offered no advantages over more conventional projection methods and their examination was therefore not contin- ued. Briefly, both methods presented a single elevation of the object which acted as the key to show the sequence of cross-section views which accompanied the key view. It was thought at the time that if successful, this sequenced cross- section technique might be applicable to complicated objects with interior as well as exterior detail. It hardly needs mentioning that this system would increase drawing time. Hence in the absence of clear cut gains in user performance there was little encouragement to pursue this technique at this stage. If experiments were performed involving shapes with interior details the results might vindicate this method.

The differences between the different types of user were less than expected. Draughtsmen are generally faster and

tend to achieve a higher proportion of correct assemblies, but it is doubtful if the differences reach statistical significance. Of greater interest in this context is whether inspectors and machine shop operatives performed better with Third Angle drawings than with First Angle drawings. The differences were marginal and not very consistent: both inspectors and machine shop operatives found Third Angle gave quicker assembly times, but whereas inspectors found First Angle better than Third for correctness of assembly, machine shop operatives preferred Third. Draughtsmen favoured Third Angle for assembly time, First for accuracy. Performances on truncated spherical assemblies were not examined statistically because it is clear by inspection of Table 2 that a similar rather inconsistent pattern of results would be obtained.

Effects of age

The ages of the draughtsmen ranged from 22 to 62 years so it was of interest to examine whether performance and age were related. The Spearman correlation coefficients were not significant either for overall median assembly times or numbers of errors made on all projections combined.The tendency was for time taken and error incidence to increase with age. The ages of the inspectors and machine shop operatives ranged from 22 years to 62 years. Corre- lation between age and overall median assembly time was insignificant but between age and overall number of assembly errors there was a marginally significant relation indicating that older men tended to make more errors. Two alternative inferences may be made from these results. First, if it is assumed that all the men studied had spent all their working lives doing the same work, then the benefits of experience seem to nicely balance possible negative effects of age to produce the low observed correlations. Secondly, if it is assumed that most of the men were comp- aratively new to the job being done when they were studied, then it seems that age has little serious influence on performance.

Table 4 Assembly performances with square prism and spherical objects com pared for draughtsmen on three projection systems

Object

Projection

1st angle 3rd angle Isometric

Median % age Median % age Median % age assembly Correct assembly Correct assembly Correct time assembly time assembly time assembly

Square prism

Sphere

Statistical significance of Time Diff- erence and error proportion

7' 0" 33.3% 4'37" 37.0% 1'37" 85.0°/0

4'36" 89.0°/0 2'28" 57.0°/0 3'29" 16.0%

Not sig. Sig: p<'002 Sig: p=-01 Not sig. Sig: p='02 Sig : n<'001


A final obvious question that can be asked of the results is whether there is a relation between time taken and errors made. Neither for draughtsmen nor for inspectors and machine shop operatives is there a significant relationship although there is a tendency for shorter times to be associated with fewer errors.

Conclusions on shape comprehension

What do these experiments tell us about how to lay out a drawing so that the shape of the drawn object shall be comprehensible?

First, it seems likely that any lay-out which clearly conveys the principle underlying its lay-out will be most effective in terms both of time taken and mistakes made in reading it. Secondly, this double bonus seems most likely to result from a Third Angle principle in the arrangement of views. Thirdly, as suggested earlier (Spencer, 1963), t,oo many rather than too few views are to be preferred. In BS 308: (1964), on page 19 the problem of the number of views is dealt with asfollows:

Number of views. The number of views should be the minimum necessary to ensure that there will be no mis- understanding. Views should be selected to give as few hidden lines as possible. Scrap or local views may be used to define particular features where a complete view of the object is not necessary.

It is suggested that BS 308 (1964) is implying that the number of views should correspond with the minimum logically necessary to convey all the shape information. Here we are saying that more than the minimum number is better, without defining how much redundancy is desirable. For example, a high degree of redundancy is desirable in the early stages of skill learning and under conditions which hamper the transmission of information. Drawings of objects familiar to production personnel in a factory need less redundancy than drawings of unfamiliar objects which these same people may be required to manufacture for the first time.

Fourthly it seems likely that views which maximise the representational appearance of the drawn object are preferable for the reader although more expensive in draughting time. Consider for example how many puzzle photographs of familiar objects are simply orthogonal views of objects normally viewed obliquely.

Fifthly, the technique used to study behaviour described in this paper is simple enough to be applied to many specific situations in drawing offices where there may be a

division of opinion on some point of drawing lay-out. Frequently, the opinion which is most obstinately defended, is not necessarily the best or the only answer. Measurement and close observation of how people actually use drawn information is capable of providing answers which are more soundly based. The important points to watch when carrying out such a study are, very briefly:

1. Use a number of subjects, chosen from the relevant group whose work will be affected by the changes in drawing lay-out.

2. Choose an experimental task which has reasonable validity but yet permits 'accurate' measurement.

3. Try to establish the statistical significance that attaches to any differences found. This means design- ing the experiment to avoid biasses (eg due to order of problem presentation) and with a sufficiently large sample of subjects to obtain re.asonable estimates of means, variances and so forth.

Acknowledgement The work described in this article was financed by a

grant from the Science Research Council.

References

Booker, PJ. 1963 I~ history of engineering drawing'. Chatto and Windus. British Standards Institution 1944 BS 1100 Part 9. Office aids to the factory. Available

from British Institute of Management, Management House, 80 Setter Lane, London EC4

British Standards Institution 1964 BS 308. Engineering drawing practice. Spencer, J. 1963 Occup. Psychol., 37,181-195. A preliminary

enquiry into engineering drawing comprehension. Spencer, J. 1965 Ergonomics, 8, 93-110. Experiments on engineering

drawing comprehension. Turner, B. 1967 Engineering Designer, September. Effective communi-

cation.

© J. Spencer and N.D. Milnes-Walker 1971


Documents

The ergonomic study of engineering drawings