Typographic measurement

A critique and a proposal

Séamas Ó Brógáin

This is a revised version (completed in September 2006) of the paper “Typo­graphic measure­ment: A critique and a proposal,” published in Profes­sional Printer: Journal of the Insti­tute of Print­ing, vol. 27, no. 5 (1983), p. 9–14. It is repro­duced here with the per­mis­sion of the Insti­tute of Paper, Print­ing and Publishing.

 Parts of this page will not display properly in browsers that do not imple­ment current stan­dards for mathe­mati­cal mark-​up, including Internet Explorer, Safari, Opera, and Chromium. Browsers that work correctly include Firefox, Netscape, and Camino.

Introduction

First introduced more than two hundred years ago, the point system has been uni­versally used for the last hundred years for express­ing the size of type. With the adop­tion of personal com­puters from the 1980s, this sys­tem has been carried over to computer-​assisted editing, type­set­ting, and design. From its incep­tion, how­ever, it has had two sig­nifi­cant drawbacks:

 

The continued use into the com­puter era of this doubly in­coherent sys­tem makes for par­ticu­lar diffi­culties in design and page layout. With the “body size” system generally, the object nominally measured—in whatever units—is not visible and cannot usually be measured on the printed page; with digital type it has no real exist­ence. At the same time the designer, typi­cally work­ing in milli­metres, is com­pelled to mix milli­metres with points in order to specify type size or line-spacing.

Postscript fonts are typically created on a grid 1,000 units deep, and Truetype fonts on a grid 2,048 units deep, within which the type­face designer chooses where to place the base­line and mean line, thus establish­ing the relation­ship between nomi­nal size and x height for that type­face. (The newer Opentype format contains either Postscript or Truetype out­lines, and the proposed Open Font format is in turn based on Opentype.) When type is used in an appli­ca­tion it is the whole of this grid that is given the nomi­nal (point) size.

A brief history

 For hundreds of years, type­set­ting involved the manipu­la­tion of lead type—small blocks of lead with the image moulded on one end—which were picked up, one for each char­ac­ter, and assembled by hand. Each piece of type was a solid object with a measurable ver­ti­cal dimen­sion; and it was this “body size” that acted as nomi­nal type size. From the end of the nine­teenth century, with the inven­tion of the Monotype and Linotype machines, this pro­cess was mecha­nised, but it still pro­duced lead type (either single char­ac­ters or com­plete lines) of a fixed depth.

Today there is virtually no lead type; the dozen or so familiar book types of the early twentieth century have given way to thousands of digi­tal types; and the print­ing crafts­man has been replaced by a graphic designer who has never seen lead type and may not even know what the measure­ment given in points refers to. (It is often stated that it is a measure­ment of the over­all size—the distance between the ascender line and the descender line—which is not correct.)

From the nomi­nal size of metal type, printers and their customers referred to the printed image as “12-point type,” “10-point type,” and so on. There is no part of the printed image that can be measured and said to be 12 points or 10 points, though printers and others familiar with the commonly used type­faces had a reason­able idea of what they would look like in differ­ent sizes. The measure­ment from one base­line to the next will pro­duce this figure, but only if the type is set solid, i.e. if there is no “leading” or additional line-​spacing, but this is some­thing that cannot be assumed; and it still leaves us with a dimen­sion not con­nected except in a very loose way with the visual size.

For the printer the most important con­sider­a­tion was that the differ­ent pieces of lead type should fit together exactly, regard­less of the design of the type or the foundry in which it had been cast. It was this that pro­vided the impetus for the develop­ment of the point system when printers began to buy type from differ­ent type­founders, rather than cast­ing their own.

A number of French inventors in the eighteenth century had the idea of a measure­ment system using a very small unit that could be the basis of a range of related sizes. All but the first (Sébastien Truchet in 1725) used one seventy-​second of the con­tem­por­ary inch—a reasonable choice for a duo­decimal world. The inch, however, has had slightly differ­ent dimen­sions in differ­ent countries and at differ­ent times, and so there­fore has the typo­graphic point.

The unit proposed by the punch-​cutter and type­founder Pierre-​Simon Fournier (Fournier le Jeune) from 1737 was one seventy-​second of the con­tem­por­ary French inch, though this was not legally defined. The point was more precisely defined by the printer François-​Ambroise Didot from about 1783 as 1864 of the pied du roi or royal foot, and this unit (≈ 0.376 mm), now generally known as the Didot point, came to be used generally through­out Con­ti­nen­tal Europe.1

The American point (generally called the Anglo-​American point follow­ing its adop­tion in England from the end of the nine­teenth century) was defined in 1886 by the Type Founders’ Associ­ation of the United States as 35996 cm (≈ 0.351 mm).2 The Monotype Corporation used its own defi­nition of the Anglo-​American point for its widely used type­cast­ing machines (to facili­tate its ingenious “unit” system), defined as 0.0007685 inch × 18 = 0.013833 inch (≈ 0.351 mm also).3

At the begin­ning of the com­puter era, the point used by the type­set­ting language Tex (from about 1978) was defined as 172.27 inch, a novel formu­lation that pro­duces a figure almost identi­cal to the Anglo-​American and Monotype points. With the intro­duc­tion of the first Apple Macintosh com­puter in 1984 a monitor was provided that had exactly 72 pixels per inch as well as a display system that showed text in various type­faces and at actual size (“what you see is what you get”), so intro­ducing the com­puter point, of exactly one seventy-​second of a stan­dard inch (25.4 mm) or approxi­mately 0.353 mm—the point now generally used in digi­tal typog­raphy. Because the Postscript page descrip­tion language was intro­duced shortly after­wards, this unit is also called the “Postscript point.”

Proposals for reform

There have been two false starts in the reform of the point system. When the first modern efforts to “metri­cate” typog­raphy were made, the pro­posers did just that: they con­verted the dimen­sions of lead type from points to milli­metres, and generally also rational­ised the range of sizes. The most notable proposal was that of the Dutch printer Adolf Stork—later director of the Stads­drukkerij, Amsterdam—whose scheme was imple­mented in a number of Dutch, French and German print­ing houses.4 A simi­lar approach was adopted by the British Feder­a­tion of Master Printers, result­ing in a British stan­dard, BS 4786 (1972), which was simul­taneously sub­mitted to the Inter­national Organi­za­tion for Stan­dard­ization as a possible inter­national standard; but it was not imple­mented any­where and was later withdrawn.5

National standards were subse­quently adopted (or con­firmed) in some countries that dis­regarded the develop­ing con­sensus that some aspect of the visual image should be the basis of nomi­nal type size. A Japanese stan­dard dating from 1962 and con­firmed in 2006 specifies the use of the milli­metre (alongside the American point, which it defines as 0.3514 mm), though it also intro­duces a pseudo-​unit, the Q (for “quarter”), as a special name for 0.25 mm.6 (By “pseudo-​unit” I mean a special name given to what is merely a frac­tion of an exist­ing stan­dard unit.) Such standards, by specifying dimen­sions in milli­metres, may be per­ceived as elimi­nat­ing the use of the point but are in fact irrelevant, as designers, type­setters and other com­puter users in all countries use the same com­puters and appli­ca­tions and are thereby locked in to the com­puter point.

The other proposed reform, and the first based on a visual aspect, was pioneered from about 1966 by the Austrian-​British graphic designer and teacher Ernest Hoch, originally under the auspices of the Inter­national Council of Graphic Design Associ­ations (ICOGRADA). Hoch and his col­leagues proposed the height of the capi­tal letters in milli­metres, called H-height, as the nomi­nal size.7 (Capi­tal height is normally the same as ascender height, though in some type­face designs it is a little less.) Though this was a break­through, bring­ing in for the first time the con­cept of a measur­able dimen­sion of the printed char­ac­ter, it still failed to indi­cate the true visual size while at the same time it alienated those who wished to retain mechani­cal (body) size. Per­haps an expla­na­tion for this false start is to be found in the historical priority and sig­nifi­cance of capi­tal letters. For the pur­poses of measure­ment, how­ever, what we are con­cerned with is the selec­tion of a single dimen­sion to act as the nomi­nal size of type and as the basis for equat­ing and align­ing differ­ent type­faces. No histori­cal or cultural judge­ments are involved: it is simply a measure­ment of size.

In 1975 the International Organi­za­tion for Stan­dar­di­za­tion estab­lished a “Typo­graphic Measure­ment” work­ing group, of which Hoch was appointed con­vener, and this became the battle­ground between those who wished to push ahead for a visual dimen­sion and those who wished to retain body size—under a variety of names—despite the rapidly changing tech­nology.8 Draft stan­dards were drawn up, based on the H-height approach, but agree­ment could not be reached, and the project was sus­pended in 1982 and formally ceased in 1984.

The revised German standard of 1999 adopts the approach aban­doned in the British stan­dards pro­cess by specifying schriftgröße (“type size,” i.e. body size) in milli­metres but goes further by pro­pos­ing that the manu­fac­turers of fonts make capi­tal height a fixed frac­tion of this size for all type­face designs, namely 66.7 per cent (i.e. 23) for the Didot point system or 70.9 per cent for the Anglo-​American point system—though com­puter appli­cations use neither of these points but rather the com­puter point.9 This is clearly an attempt at recon­cil­ing the con­flict­ing require­ments of a system based on the now non-​existent body size and one based on some visual aspect. A general imple­men­ta­tion of this system would require that all type­face designers and font manu­fac­turers apply this formula to newly manu­fac­tured fonts; and it could not be applied to the millions of copies of the thousands of fonts already in use.

Special unit, or special name? Despite the weight of tradition, there is no tech­ni­cal need for any special measure­ment unit in typog­raphy, with or without a special name. The 16th General Con­fer­ence on Weights and Measures (1979) resolved “that the pro­lifera­tion of special names repre­sents a danger for the Système Inter­national d’Unités and must be avoided in every pos­sible way . . .”10 What we are deal­ing with is a simple linear measure­ment. Type size and spacing can be measured perfectly well in milli­metres, the same unit used for the other dimen­sions on a page.

A new approach

Numerous writers over the years have had to resort to com­plex expla­na­tions to accom­mo­date the fact that what is called the “size” of type is not really the size at all. The follow­ing example is from the classic Intro­duc­tion to Typog­raphy by Oliver Simon:

A further examination of [speci­mens of twenty-​three typefaces] shows very con­sider­able differ­ences in the size of the actual type face on the 12-pt. body . . . Plantin and Times, for instance, are, in rela­tion to others, so large on their bodies that whilst 12 pt. is the norm for almost all the types exhibited, the equiva­lent could well be 11 pt. in their case . . . This ‘large­ness’ on the body is measured first in terms of the height of the non-​ascending and non-​descending lower-​case letters, e.g. x (hence the term ‘x-height’), and secondly, the width of the m (which determines the ‘set’).11

Where objective comparisons of sizes were required, a small number of writers accepted the logical con­clusion and based these explicitly on the x height, as for example:

The typefaces of the various systems are differ­ent in their image size for the same type size . . . In order to provide an objec­tive com­pari­son, the lower-​case letters have been brought to a unified dimension.12

One area in which an objec­tive com­pari­son of type size has a par­ticu­lar impor­tance is the measure­ment of legibility (also applied in opticians’ reading tests). Unlike previous studies, which had used only a com­pari­son of “point sizes” and were there­fore question­able, the Swedish researcher Bror Zachrisson employed “visual size,” which he defined as “x-height measured in mm multiplied by the mean width of the letters.”13 (A more exact definition of visual size would certainly have to take into account the width of the letters, but this has been decided before­hand by the type designer; the only dimen­sion under the control of the user of a type­face is the vertical one, with the hori­zontal dimen­sion changing accordingly.)

In the illustration below, the first word in each line (in Garamont) is 24 pt, while the second word (in Helvetica) is 24 pt in the first line, 2212 pt in the second line, and 18 pt in the third line. Accord­ing to the present system, the two words in the first line are the same size. Clearly, how­ever, they are not.

 

Using capi­tal height as nominal size, the two words in the second line should be equated. Though this is an improve­ment, it is hard to see how it could be regarded as an acceptable solution.

One has only to scan any page of text (or the text of this web page) to see that what con­sti­tutes the text is rows of vary­ing char­ac­ter shapes that, in com­bi­nation, present to the eye a quite uniform pattern—a pattern created by the x height. The pro­trud­ing ascenders and descenders, and occasional capi­tal letters, are not sufficient to break this pattern. Here, surely, is the “size” of the type. The con­clu­sion to be drawn is that x height, given in milli­metres, should be used for describ­ing and specify­ing type size. The con­cept of a stan­dard­ised range of sizes is redun­dant, as digi­tal type can be scaled up or down to any size.

It is necessary to accommodate two other dimen­sions: line-​spacing and capital height. Though the former may cor­res­pond to the histori­cal body size, and the latter cor­res­ponds to the dimen­sion chosen for the earlier reform propo­sal, they are included here for objec­tive reasons, in that it may be neces­sary to specify them inde­pen­dently. Because of the strength of the argu­ment in favour of the x height as nomi­nal size, these should be regarded as minor com­pli­ca­tions rather than as counter-arguments.

Line-spacing. The specification for type size combined with line-​spacing has traditionally used a simple nota­tion, namely the two figures separ­ated by an oblique stroke, as for example 10/12 pt (or simply 10/12). An increase in line-​spacing was originally achieved by adding thin strips of lead between the lines and later by casting (for example) 10-point type on a 12-point body, as a result of which this nota­tion is verbal­ised as “ten on twelve (point).” There is no reason not to use the same nota­tion when giving dimen­sions in milli­metres, as for example 1.5/4.

Capital letters. There are certain circum­stances in which capi­tal letters are used alone, with­out their cor­res­pond­ing lower case (or with­out neces­sarily having a lower case). An un­ambiguous solu­tion would be to enclose the figure for capi­tal height in paren­theses; and this too could be com­bined if neces­sary with the figure for line-​spacing, as for example (5)/7.5.

Implementation

In digital fonts it is possible for the type designer or manu­facturer to specify explicitly the dimen­sion of the x height for a par­ticu­lar type­face (in terms of the number of em units within the design grid); but whether this is done or not it is also feasible for appli­cations to be supplied with the height of the actual letter x. Un­fortunately, appli­cations would have to be re­written to allow this dimen­sion to be used as the nominal size, and for this to happen it would be necessary for a con­sensus or at least a certain demand for such a feature to have arisen. On the other hand, this feature might at first be provided as an option rather than as the default method; and it could be intro­duced piece­meal in indi­vidual appli­ca­tions, with­out waiting on any global or synchro­nised change­over. Most impor­tantly, no changes would be required to exist­ing digi­tal fonts.

Fonts in web pages. In cascading stylesheets (used in con­junc­tion with the mark-​up language for web pages) it is already pos­sible to dispense with the point, as type size (and line-​spacing) can be speci­fied in milli­metres, thus elimi­nat­ing one half of the problem. What is specified, how­ever, is the em square, defined as “an abstract square whose height is the intended distance between lines of type in the same type size”—in other words, the historical body size.14 This is a non-​existent dimen­sion as far as the user is concerned.

It is gratifying to note in the same speci­fi­ca­tion the state­ment that “the sub­jec­tive apparent size and legi­bility of a font are less dependent on their ‘font-​size’ value than on the value of their ‘x-height’, or, more use­fully, on the ratio of these two values,” for which the term aspect value is intro­duced. How­ever, the ability to specify type size in terms of this value is not offered.

Application to other scripts

The system described above, based on the require­ments of the Roman alphabet, can be applied to other scripts also, which for this pur­pose it is con­venient to divide into three categories.

1. The Greek and Cyrillic alpha­bets have a structure simi­lar to the Roman alpha­bet in having distinct capi­tal and lower-​case letters.

 

2. Chinese, Japanese and Korean characters, being essentially square and of a uni­form height, as well as relatively complex, lend them­selves to being equated with capi­tal letters for the purposes of align­ment and measure­ment. Traditional publish­ing prac­tice has generally been to centre such char­ac­ters vertically (i.e. between base­lines), though com­puter appli­ca­tions, includ­ing web browsers, normally use the base-​alignment model.

3. Other scripts—including Arabic, Hebrew, and the scripts of India and south-​east Asia—require separ­ate decisions, based on expert knowledge; but the criterion in each case should be the same: for the nomi­nal size a dimen­sion should be chosen that best repre­sents the visual size. This could be aligned when neces­sary with the x height (or, exceptionally, the capi­tal height) of the Roman alphabet.

References

  1. Andrew Boag, “Typographic measurement: A chronology,” Typog­raphy Papers, 1, Reading: Depart­ment of Typog­raphy and Graphic Com­muni­ca­tion, University of Reading, 1996, p. 105–121.
  2. Richard L. Hopkins, Origin of the American Point System for Printers’ Type Measure­ment, Terra Alta (W.Va.): Hill and Dale Press, 1976.
  3. Monotype Corporation, A Monotype Com­pos­ing Machine Described: For Pros­pec­tive Users, Over­seers and Students, London: Monotype Cor­por­a­tion [1953].
  4. A. Stork, De Invoering van een Metrisch Maatsysteem in de Typografie, Amsterdam: Verenigde Drukkerij Dico, 1946; also A. Stork, The Intro­duc­tion of a Metric System in Typography [address to the Eighth Inter­national Congress of Print­ing and Allied Industries, Venezia, 1954], Amsterdam: [A. Stork], 1967.
  5. British Standards Institution, British Standard 4786 (1972): Specifi­cation for Metric Typo­graphic Measure­ment, London: BSI, 1972; with­drawn 1985.
  6. 日本工業標準調査会 [Japanese Industrial Standards Committee], JIS Z 8305 (1962): 活字の基準寸法 [Dimen­sions of Print­ing Types], Tôkyô: 日本規格協会 [Japanese Stan­dards Associ­ation], 1962; con­firmed 2006.
  7. E. Hoch and M. Goldring, “Type size: A system of dimensional references,” Typographica, 13 (June 1966); also Ernest Hoch, “Typo­graphic metri­ca­tion,” in Penrose: Inter­national Review of the Graphic Arts, vol. 70, London: Northwood Publi­ca­tions, 1977, p. 24–26.
  8. Private correspondence with the late Ernest Hoch.
  9. Deutsches Institut für Normung [German Institute for Standardisation], Deutsche Norm DIN 16507-2: Drucktechnik—Schriftgrößen—Teil 2: Digitaler Satz und Verwandte Techniken [German Standard 16507-2: Printing Technology—Type Sizes—Part 2: Digital Type­set­ting and Related Techniques], Berlin: Beuth Verlag, 1999.
  10. Bureau International des Poids et Mesures, Comptes Rendus de la 16e Conférence Générale des Poids et Mesures (1979), Sèvres: BIPM, 1980, Reso­lu­tion 5, p. 100.
  11. Oliver Simon (edited by David Bland), Introduction to Typography, London: Faber and Faber, 1963, p. 18.
  12. Roland Zimmermann and Heinrich Fleischhacker, Balancing Type Widths, and the Different Effects of Metal, Photo, CRT and Laser Setting Systems (supplement to Typografische Monats­blätter, no. 5, 1982), p. 16.
  13. Bror Zachrisson, Studies in the Legibility of Printed Text, Stockholm: Almqvist och Wiksell, 1965.
  14. World Wide Web Consortium, “Cascading Style Sheets, Level 2: CSS2 Specification [1998],” at http://www.w3.org/TR/REC-CSS2.

Postscript

In March 2013 Dr Stanley Max of Towson University (Towson, Maryland) suggested using the micrometre (one-​thousandth of a millimetre) as the unit of measure­ment for nomi­nal type size. This would elimi­nate the need for deci­mal frac­tions; at the same time it gives much larger numbers, and these convey a precision that is not visually per­cep­tible. For example, the size given above as 1.5 would be 1500, and the example including line-​spacing, 1.5/4, would be 1500/4000.

Comments to

 Home page

Made with GNU/Linux        Made with Unicode encoding        XHTML validation