john Gill technology header image

The Tiresias Family of Fonts

Dr. John Gill
Dr. Sylvie Perera
July 2003

The design of a typeface can significantly affect legibility and readability. For many applications the aesthetic appearance of the text is the basis on which a font is selected. However there are occasions when legibility is the paramount consideration (eg for notices to be read in the event of a fire).

In the ideal world, the user should be able to select their preferred typeface, size, line spacing and line length. This approach is viable when producing in small quantities (eg school books for partially sighted children) or the information is displayed on a computer screen.

On a personal computer, the user can set a number of parameters to suit their individual needs and store these as the default mode. On a public terminal, such as a cash dispenser, this approach is not possible but the user's preferred presentation could be stored on their card; there is a European standard for coding such information on the user's card.

With text broadcast for display on television sets, the layout has to be specified before being broadcast since the current television sets have very limited processing power. The change from analogue to digital broadcasting offered the opportunity to use a more legible font than the mosaic one on analogue television.

Tiresias Screenfont was developed specifically for subtitling on UK digital terrestrial television. The character shapes were designed to help differentiation of one character from another. For people with impaired vision, or inappropriate spectacles, some numerals such as 6, 8 and 9 can look very similar. Screenfont has open shapes, designed to make each character as clear as possible. Characters such as the lower case 'l' have been designed to ensure they are different from the numeral '1' or an upper case 'I'. These features have now become particularly important with the growing use of unusual combinations of letters and numerals in email addresses and passwords.

The development process involved experimental studies on character shapes, the relative weight of characters, inter-character spacing and optimum size given the limitations of presentation on a television screen. The broadcasters provided video tapes of television programmes subtitled with a range of experimental fonts so that the various options could be evaluated with a cross-section of potential users.

Screenfont was originally designed with 180 characters, but this has now been extended to 350 characters to permit the representation of forty European languages.

The technology for rendering a font on a television set is different than that for computer screens. Therefore Tiresias PCfont was developed for computer terminals such as cash dispensers, ticket selling machines and information kiosks. PCfont is basically similar in appearance to Screenfont. However this font has been found useful by some partially sighted computer users.

Many information labels on public terminals are difficult to read, particularly for someone wearing bifocal spectacles. Tiresias InfoFont was designed for reading at distances from 30 to 100 cm. The font is similar but bolder than PCfont, but the character shapes have been adjusted to maintain legibility. This font has been used for labelling exhibits in museums, and for fire notices.

The work on these fonts has prompted enquiries from organisations wanting to ensure that legibility is improved in other areas such as signs in buildings and public spaces. Tiresias SignFont is a bolder typeface, but has characters that have designed to maintain open shapes that provide maximum readability at longer distances.

All these fonts have been developed for applications where a single font has to employed which is legible for as many people as is reasonably possible. This is an example of inclusive design. However there are also applications where the user population is a special group within society and their needs may be significantly different.

Many older people have difficulty in reading standard text even with appropriate spectacles and optimum illumination. The situation is made worse by the use of small characters with fine strokes, printed on a patterned background. Also many people who can read standard text find it very tiring, so prefer to read large print when they can. Large print publications should be designed to specifically help with reading problems, and should not just be an enlarged version of the ordinary print.

The design of publications in large print for people with low vision is dependent on the specific target audience. The majority of large print readers appear to be over retirement age (Bell, 1980). Many older people have difficulty reading standard text even with appropriate spectacles and optimum illumination. Also many people who can read standard text find it tiring, so prefer to read large print when they can. Most large print books in the UK are purchased by public libraries for older readers who may have macular degeneration or cataracts. However the choice of typeface used in large print publications, does not appear to have been based on scientific research.

The ageing process significantly affects the human eye. Less light reaches the retina, and a decrease in the ability of the eye to change its focus produces an unclear image on the retina. Also eye conditions such as macular degeneration and cataracts are age-related so these conditions are becoming more common.

Macular degeneration is the main cause of low vision in Europe. The macula is at the centre of the retina and is used for detailed activities such as reading and detecting colours. People with macular degeneration have to use the less sensitive areas of the retina for recognising characters. This has significant implications on the design of a font for large print publications.

The research involved a review of the relevant literature, interviewing publishers of large print materials and others involved in delivering large print materials to the readers. Then a series of practical experiments were undertaken with existing and potential adult readers of publications in large print. The design project independently examined parameters including:

  • Characters' shape and weight
  • Letter and word spacing
  • The effectiveness of serifs
  • The shape and size of punctuation

Due to traditional printing techniques, typefaces are measured in point sizes - one inch equals a point size of seventy-two. This represents the vertical size of the metal cast even though the character size on this face may vary, so printed typefaces with the same point size can differ in height. In this project the x-height will be used so typefaces can be compared. The x-height is the height of lower case letters such as the 'x' and 'o'. The taller the x-height of the characters, the more white space that can be incorporated into the body of the letter thereby improving the legibility (White, 1987).

Due to size constraints of large print books, large print producers had researched the most appropriate size of print for use in their books and had found 16 point to be suitable. Although size is a crucial factor for VIPs ability to read, within this investigation all typefaces tested had an x-height equivalent to 16 point Times New Roman.

'Legibility' as defined by Fabian (2000) is a measure of the recognisability of the characters, how easily they can be read, based on the visual appearance in a given environment. Vernon (1931) found confusable letters include f & t, l & t, c & e, n & a, i & j, I & J. Burt (1974) maintains that distinguish ability is necessary between I l 1 ! i and h & b, C & G, Q & O, J & F, R & Q. For numerals Tinker (1928) found 3 5 8 2 to have low legibility.

Initially, to determine if currently used typefaces - Times New Roman and Arial - are regarded as the most legible by large print readers, they were tested against each other. So for the first part of the investigation to find the 'most legible typeface', the null hypothesis stated that there will be no significant difference in legibility between the typefaces. The second hypothesis related to the 'degree of serif'. It was thought that for visually impaired readers, a sans serif typeface is more legible than a seriffed one. It was also thought that the legibility will decrease with an increase in the degree of serif. Hypothesis three about the 'inter-character spacing' stated that the wider the spacing the more legible the typeface for visually impaired readers. The fourth hypothesis in relation to the 'weight' of the typeface assumed that the bolder the typeface the more legible it is for visually impaired large print readers. This was thought to be due to an increase in contrast which is especially important for older readers. The fifth hypothesis as presented by previous research stated that the more pronounced the 'punctuation marks', the more legible visually impaired large print readers will find them.

Potential and actual adult readers of large print books from a variety of sources around the UK participated in the research. A variety of visual impairments are prevalent amongst large print book readers and in many cases a reader may have more than one disorder. This investigation regarded visual impairments generally without specific reference to a particular disorder. This was based on the assumption that regardless of the individual's impairment, the print should be suitable for any reader.

Over 4000 fonts were examined to see if any were appropriate as a legible typeface for a large print book. Typefaces used by large print producers, large print magazines and Tiresias were also compared. It was decided that current large print typefaces and Tiresias should be investigated for legibility. Using Fontographer 4.1, Tiresias was decreased in weight to be comparable with the other typefaces.

The experiment to determine the most legible of 3 typefaces was a within-subjects design. The independent variable was the typeface and the dependent variable was its legibility. The subjective factor of ease is probably of great importance in reading (Zachrisson, 1965), readers opinions of accessibility are relevant measures of legibility. Subjects responded about the legibility of the typefaces using a questionnaire (written in 16pt Verdana) and a sample of continuous text replicated three times (in Times New Roman, Arial and Tiresias).

A two page extract was taken from a popular author's large print book. The samples were formatted as if they were an open double page in a large print book, with the sample name at the top e.g. A, One and Alpha. Variables were used as close to their optimum as possible and the general format and constraints of a large print book were also adopted e.g. black ink on white paper was used to provide a high contrast. Only left justification was used. No italics or all-caps were used. The alignment of paragraphs/indents/margins were in large print book format with 50-65 characters per line. The margins for all the samples were the same. The line spacing was adjusted so all the samples required the same amount of vertical space on the pages. Therefore the only variable was the typeface.

A succinct questionnaire was designed to request the subjective legibility rating and preferred readability of the typefaces. A three point scale (used to make answering easier for visually impaired people) allowed for absolute rather than relative rating. Further questions inquired as to the preferred and least preferred typeface for use in a large print book. The order of the samples was altered for each questionnaire so 6 versions of the questionnaire were produced and distributed in equal numbers in a random order. A pilot of the samples and questionnaire revealed no major difficulties with the questionnaire design although some questions were clarified.

The graph below indicates that Tiresias was seen as 'very legible' by more subjects than either of the other typefaces. Tiresias was also regarded as 'not legible' by fewer subjects than the other two typefaces. More people found Tiresias 'OK' or 'very legible' compared to the other two typefaces. Arial appears to be more accepted than Times New Roman.

The data was ordinal with 'typeface' as the only variable, because this variable had three levels and the same subjects in each group, a Friedman test was performed on the data after the ratings were converted to ranks. Resulting large Chi-squared values suggest the null hypothesis is not true, X2 = 20.26, df = 2, p<0.0005. There was a significant difference between the legibility of the typefaces, so to find out where this significance lay, Wilcoxon Related tests were performed between the levels.

The mean ranks for each typeface:

Mean rank

For the Wilcoxon test between Arial and Tiresias, z = 2.80, p <0.01, the difference was significant. From the table above, it is evident that Tiresias had a higher rank than Arial, therefore was more legible. For the Wilcoxon test between Times New Roman and Tiresias, z = 3.40, p<0.0005, this difference was also significant. From the table above, it is evident that Tiresias had a higher rank and was more legible than Times New Roman. Between Times New Roman and Arial the Wilcoxon results were, z = 1.75, p>0.05, this reveals there was not a significant difference between these typefaces.

Tiresias was shown to be the preferred typeface for use in a large print book and statistically shown to be significantly more legible than Arial or Times New Roman. The majority of subjects also preferred to read a large print book in Tiresias rather than the other typefaces, this was especially evident for those with poor vision. Tiresias was designed for legibility and is concurrent with Hill's (1999) statement that the form of the letters contributes to its legibility.

As people's level of reading vision improved, their typeface preference was unclear so other factors may have been influencing their choice of typeface. Subjects least preferred typeface for use in a large print book was Times New Roman, which is the equivalent of some typefaces currently used.

Further studies of specific typographic factors provided indications of the preferred levels by people that were current or potential readers of large print books. The results revealed that people with low vision are not a homogeneous population so an 'optimised' typeface may not apply to every reader. Despite this, a large print version of Tiresias was designed and developed that was regarded as more legible than those currently employed.

The typeface with the darkest weight that was tested, was favoured by the subjects. This supports the fourth hypothesis which states that 'the bolder the typeface the more legible it is to visually impaired large print readers'. The preferred typeface is darker than currently used typefaces in large print books.

Normal spacing, similar to that used in large print books was preferred and had a higher reading speed than the enlarged spacing tested. This finding is contrary to the third hypothesis that was based on the belief that the wider the spacing, the more legible the typeface would be for visually impaired readers.

Burt (1974) discloses three main types of serif:

It is thought that hairline serifs are equivalent to hair-line strokes which are not beneficial to partially sighted people. Slab serifs are full, solid serifs. Bracketed serifs accentuate the end of the stroke.

The semi-serif version of the typeface that resembled a bracketed serif (which is not as exaggerated as a slab serif) produced a higher reading speed than either a sans serif or seriffed version of the typeface. The degree of serif hypothesis was not supported. It appears that a slight degree of serif which accentuates the characters' ends without distracting from the simple form actually increased legibility. There is not a linear correlation between the degree of serif and legibility.

Punctuation marks that were rounded and slightly enlarged were preferred by subjects. This partially supports the fifth hypothesis, the punctuation marks were more legible when exaggerated but not at the point where they were too bold and distracted from the text.

Based on this research, LPfont was designed to maximise character definition and legibility for readers of large print publications. The font is now generally available with a character set accommodating forty different European languages. A full scientific report of the research involved in its development is available at report/index.htm


Burt, C. (1974) A psychological study of typography. Cambridge: Cambridge University Press.

Fabian, N. (2000) Legibility of typefaces: ending the confusion.

Hill, B. (1999) The magic of reading.

Vernon, M. D. (1931) The experimental study of reading. Cambridge: Cambridge University Press.

White, A. W. (1987) How to spec type. New York: Watson-Guptill Publications.

Zachrisson, B. (1965) Studies in the legibility of printed text. Uppsala: Almquist and Wiksell.


John Gill Technology Limited Footer
John Gill Technology Limited Footer