This article is from the British Printer magazine of 1961. The research was conducted for PATRA: The Printing and Allied Trades’ Research Association.
The proportions of characters making up a fount of type should be such that by the time one character is exhausted as little as possible of the others remains in the case. A further requirement is that any character in the fount has the same chance of being exhausted first and still leave a nearly empty case. If this is achieved then the printer will effect savings in the amount of type stocked in the cases and also in the expensive reordering of individual characters to make up his deficiencies. This article presents the results of the first systematic study of type fount proportions and a new scheme is proposed which it is believed will fulfil the above requirements, as far as it is practicable.
Such a study may seem a little belated, but hand-set work still remains an important part of printing. It is estimated, for example, that the fount schemes proposed in this article will effect at least a 10 per cent saving of dead metal in the case which to the industry represents many thousands of tons of type metal. Furthermore, as the kind of work which is now hand-set has become somewhat stabilised the fount proportions proposed should remain effective for many years to come.
The origin of type fount proportions, even in recent times, is rather obscure and this seems largely because the responsibility for supply is confined to relatively few people. It is certain, however, that as early as the beginning of the sixteenth century some account was being taken of the variation in usage of the various characters since the most frequently used characters were placed at the front of the case. A more positive example is given by Moxon’s lowercase, which appeared in 1683, and which has remained virtually unchanged to this day. The lay of this case is such that the volume of the type compartments is roughly proportional to the frequency of usage.
Some of the books on printing which were produced in the nineteenth century contain tables of bills of founts, but unfortunately they rarely mention how these proportions were determined. It is true that the only satisfactory way in which to arrive at suitable proportions is to count the frequency of occurrence of characters in a piece of work that has been hand-set. Presumably most of them were determined in this way but without a knowledge of the nature of the work chosen, the validity of the results cannot be judged. One of the few records that do exist of a count illustrates this point. The count, which is attributed to the Caslon Foundry, was made by enumerating the number of letters used in setting a lengthy debate in the House of Commons where it was assumed that ‘the best and most comprehensive English would be spoken’. The validity of this count can be questioned on two points, firstly that the frequencies of the spoken word vary from the written word and, secondly, the sample was not typical of the English being set at that time.
It is recorded that ‘the proportions of almost every typefounder failed lamentably to give satisfaction’. Such failures seem partly due to the use of biased samples on which to base the proportions and partly to the fact that, at a time when all the work was hand-set, small variations in the style of the work would have a large effect on the characters required. The work of Dickens, for example, would quickly empty the case of vowels, whereas Macaulay’s style had a similar effect on consonants. No fount proportion scheme could reasonably be expected to cope with that type of variation.
At the present time the copy that is hand-set from roman and italic types may be broadly classed as jobbing work, and it gives rise to rather different problems than those facing the old typefounder. This change in the character of the work, which was brought about by the widespread use of typesetting machines, has led typefounders to modify the old proportions by ‘experience’ in order ‘to meet the needs of the customer’. It might be expected that since most typefounders are catering for the same type of work their experience would have led them to the same proportions. In fact, for some characters there are wide variations between the various proportion schemes in use today.
It should be noted at this stage that the present work was not undertaken as an academic exercise but the need for it was suggested by a typefounder. Subsequent enquiries amongst printers confirmed this and their main complaint was that currently used proportions gave rise to shortages of the most commonly used characters (in particular, e, r, s and t) while the least used characters built-up in the case. The reason for this happening will become apparent later.
Before the main results are discussed it is essential to realise the main types of variation that will affect the type proportions required to set a piece of jobbing work. There are three of these:
- Work-type Variation
Holiday brochures provide a good example of work-type variation since in these a consistent part of the hand-set work are the names of hotels. Consequently, the frequent occurrence of the word ‘HOTEL’ means that a higher proportion of the characters H, O, T, E, and L will be required than is normally found. This type of variation is inherent in the work.
- Job Variation
A parish magazine, for example, normally contains a large number of displayed advertisements for the particular town it serves. The frequent occurrence of the town’s name will again upset the normal proportions of characters. This variation is inherent in the job, rather than the type of work, as the characters most seriously affected will vary from town to town, ie from job to job. Furthermore, with this type of variation if a number of such jobs are undertaken for different towns then the likelihood of upsetting the normal proportions is reduced. On the other hand, with work-type variation the proportions become more seriously affected as more jobs of the same type are undertaken.
- Sampling Variation
The two types of variation denned above will upset any fount proportion scheme and this fact must be recognised by printers and catered for by separately ordering more of the characters affected. There is, however, a third type of variation which is always present and must be taken into account to the fount proportion scheme itself, This is called ‘sampling’ variation and because of its importance it is discussed in detail.
The foundation of any type fount scheme is that characters occur in fixed proportions, but the essential point is that the proportions can only be considered as fixed for a large number of characters.
To illustrate this statement, suppose that a piece of setting consists of 100 lines and each line has 50 lowercase letters. If there is no work-type or job variation present then about 200 d’s would be used in the setting. This is 4 per cent of the lowercase alphabet which is the normal proportion for d, that is, what is expected to occur in a large sample of characters such as the 5000 used in this supposed setting. If each of the 100 lines is now taken separately as small samples of 50 characters then there will not be 4 per cent, or two d’s in each line. There will be a number of lines that do not contain any d’s and it is quite possible that one line will contain as many as seven or eight. This illustrates sampling variation and shows that if only small amounts are set then a wide variation in usage is expected.
Referring still to the above example, if the occurrence of a d‑and the same can be argued for any character — is a purely random process then the probability of obtaining 0, 1, 2 etc of them in any of the lines is given by the 1st, 2nd, 3rd … terms of the binomial expansion (0.04+0.96)50. The results of this calculation are shown graphically by the full line in Figure 1, where it can be seen that with 100 lines some 13 would be expected to have no d’s, 27 have one d, 27 have two d’s and so on. The dotted line in Figure 1 shows the probabilities for samples of 25 characters, and the curve becomes more distorted and shows that the chance of getting a wider variation from the expected one d increases. Conversely, as the size of the sample is increased, so the curve becomes more symmetrical with its peak over the true proportion and the spread of the curve (the variation) getting smaller. A further fact, which is not Illustrated here, is that a character such as e, which has a higher proportional occurrence (13.4 per cent) will have a tower percentage variation for the same sample size. The value of these calculations to this study is that for a fount of a given size the number that is likely to occur for each be found.
The calculations are based, however, on the assumption that the occurrence of a character is a random process that is, its occurrence is independent of the characters previously set. This is clearly not the case when it is known that for 58 per cent of the times that d occurs r does so after n or e and that it does not normally follow letters such as c, h and j. In order to determine how this dependency would affect the calculations, a number of tests were carried out and it was found that for the present purpose of type fount proportions, the effect would be negligible. This means that the statistical model outlined above can be used to predict what variation is expected to occur under various circumstances and so place type fount proportions on a more precise basis than has hitherto been possible.
As mentioned earlier, the only way is which it is possible to determine the proportion of characters is by counting their occurrence and using this to predict future requirements. It is important when making a count to select samples of work which truly represent the type of work being hand-set at the present time and so reduce the number of characters to be counted to a reasonable level.
To develop the new scheme samples of hand-set work were obtained from twenty-five randomly-selected printing firms, which included jobbing printers, magazine printers and a provincial newspaper. In all, 92,000 characters (excluding spaces) were counted from 350 separate jobs. In order that job and work-type variations could be examined more closely these items of work were regrouped into eighty-eight classes containing jobs of a very similar nature and further regrouped into fifteen broad classes of work. These fifteen work-type groups included forms, entertainment handbills, and a variety of displayed advertisements specific to various subjects such as motoring, office equipment, chemical engineering and shop services. The characters were also subdivided into composition and display sizes, the latter being characters of 14 pt and above.
Clearly, if an examination of the various items of work showed great differences from one another, there would be no value in altering the currently used proportions. It so happened, however, that sampling variation was the variation of greatest importance. Other types of variation did occur infrequently as expected: for example, with lowercase a, two jobs that were found to show other variations were a dancing academy prospectus and a ballet programme. Some variations were not quite so obvious, such as the work-type variation shown by lowercase b which was not found so frequently as expected in displayed advertisements for shop services. The general remits of this work do show, however, that a type fount scheme which would suit most printers is entirely practicable.
The basis of the new scheme is the statistical model previously discussed. Simply interpreted this means that the less frequently used characters need to be strengthened more than the commonly occulting ones and the exact amount of strengthening can be determined mathematically. The currently used schemes also strengthen the less frequently used characters but they do so irrespective of the size of the fount and this produces excesses of these characters. By realising that when the size of the fount is increased the proportions should get closer to the actual proportions found from the counting the scheme proposed here will meet requirements of type fount proportions outlined in the introduction. An abridged version of the new founts, together with the actual proportions found is given in Table I for both lowercase and capitals. Table II shows the actual proportions found for figures and points.
|Sort||% Found||Size of Fount|
|Sort||% Found||Size of Fount|
|ACTUAL PROPORTIONS OF FIGURES AND POINTS|
|Figures, etc.||%||Points, etc.||%|
Table I is based on composition sizes but comparison of these proportions with those obtained for display sizes showed that there was little justification for suggesting separate proportions for the two groups. The main differences found were that capitals I and L occurred rather more frequently in the display sizes.
The quantity of type ordered from a typefounder varies considerably; it may be a five-pound fount for a special job or it may be sufficient to fill a case. So that the printer will get the maximum benefit from the new founts two proportion tables have been prepared. For orders less than the equivalent of 160a or I60A, which contain fewer than 2,000 characters a ‘preliminary fount’ is used which is the weighted system shown in Table I. For orders exceeding this quantity, and where the effects of sampling variation become small, the type is supplied from a ‘continuation fount’ in which the number of characters are in direct proportion to those found from the counting. This refinement, which has again been devised to give a more uniform usage from the case, will not complicate the ordering of type from the point of view of the printer.
Another aspect studied was the ratio of the number of lowercase characters to the number of capitals in a complete fount. At present a 5lb fount of jobbing type contains 2 1/2 lb of lowercase and 2 1/2 lb of capitals, figures and points. This weight relationship automatically fixes the numerical ratio and those in current use have about 1.9 lowercase for every capital. It was found, however, that a ratio of 1.5 lowercase to one capital would better suit the majority of printers and to achieve this future founts would have to be made up of 2 1/4 lb of lowercase and 2 3/4 lb of capitals, figures and points. Other ratios incorporated into the new scheme are that the most suitable ratio for capitals to figures and points is 3.8 to 1 and that of points to figures is 1.5 to 1. The latter two ratios do vary considerably with the size of the type and those suggested here are again the ones that would suit most printers.
There were many other aspects of this work which had to be discussed and studied but because of their limited interest they are not mentioned here. Nevertheless they were important in order to make the new scheme easily workable for the typefounder and also acceptable to the type user.
As quite a few firms carry out hand-setting and correcting of machine-set work from the same case it was necessary to make a further study in order to determine whether the fount scheme above would be quickly upset by such a practice. In other words, are the proportions obtained for hand-setting founts suitable for corrections founts? Eleven main reasons for corrections were listed and while some of these (batters, missing words, wrong fount, etc) would require proportions almost identical to those found for jobbing work, there were others which depended on the human element and machine capabilities. Because of the latter, no precise proportions are possible and the requirements will vary from firm to firm according to the ability of the operators and the type of work being produced. One major requirement of a corrections fount is that it must be of such a size as to withstand sudden demands made upon it as are called for by repeated mis-spelling of a word, a dirty matrix, or the replacement of one of the alphabet in the die-case by a more frequently occurring sort. If this requirement is met, then, from counts of the frequency of occurrence of characters requiring corrections the hand-set scheme produced here will prove to be quite satisfactory under most circumstances.
For suggesting the problem and providing initial evidence of its existence, I am grateful to Messrs Santype Limited. I wish especially to thank their former Managing Director, H. F. W. Cory, for his valuable help on the practical problems associated with the work.
This article from the British Printer magazine during 1961