eBooks – linear, interrupted

I’ve been happily reading G. K. Chesterton’s The man who was Thursday on my iPad. Happily, because the only print edition I own is the 1938 Penguin edition, which is set in 10/11pt Monotype Times New Roman in a style which decided pre-dates Tschichold’s Penguin composition rules: extra spaces between sentences, no word division to help justification, spaced-off punctuation, etc. Print isn’t always lovely.

Using Stanza, I could format the Project Gutenberg text pretty much as I wanted to – with indents, not line spaces to separate paragraphs, the ability to alter line feed independently of face size; yes, I managed to format quite a happy page. The iPad choice of fonts is restricted but in this case, with little italic or anything more exotic in the text, Apple/HTF’s Hoefler Text looked ok.

So why did I enjoy Thursday when I’ve previously ranted about the typographic problems with eBooks? It all comes down to the configuration of the text.


Although a fantasy (‘nightmare’ is Chesterton’s description), Thursday is pretty much a standard prose novel. A few poetry extracts (we’ll come to those later), and more than a dozen chapter headings, but otherwise just paragraphs of text. No subheadings, no pictures, no lists, no tables, no notes. Using the schema devised by Michael Twyman to describe the possible configurations of verbal graphic language, the bulk of Thursday sits firmly in the category labelled ‘linear interrupted’.

‘Linear interrupted’ means that the text just runs on, in a linear fashion, within each paragraph unit. Paragraphs themselves are ordered in a linear sequence, with no differentiation. The ‘interruption’ is only imposed by the external constraint of the frame in which the text is composed – in metal, the text area formed by the page chase, in digital page layout by the text box and its margins, or the area available on the screen in which text can be rendered. Line breaks are arbitrary, and understood by the reader as such. The arbitrariness is reinforced in traditional print by the choice of justified setting, which neatens everything up into a block, so the only deviation from a regular left or right margin is the (meaningful) white space that indicates a paragraph start or break-line.

(Of course, many typographers argue that even word spacing – even at the expense of an aligned right-hand edge – is an even better indication of linearity, through the evenness of rhythm it indicates, than the squaring up of text. But these are just two ways of indicating to the reader that the stuff just runs on.)

And even in traditional typesetting, the idea of being able to reflow a text into a given space is a common one. Books are always being reset, reformatted, larger, smaller – my Thursday is, I’m sure one of dozens of typographic presentations of that text. But as a nearly pure piece of linear interrupted, it is eminently reflowable, and fits the reflow model of the dynamic-layout eBook very well. While every book designer thinks they have hit on the perfect page for their publication, you can’t deny the need for more than one solution.

So where does the reflow model break down? Well, even in Thursday, it breaks down whenever a paragraph isn’t a ‘normal’ one. It doesn’t have to, because the text could have attributes, or the text engine rules, that prevented the following problems:

Chapter heading becoming detached from the following text. Stanza doesn’t seem to observe keep options, or maybe the headings are not properly coded as headings – after all, the one defining attribute of a heading is that it belongs with what follows, not what went before.

Poetry being ‘over-displayed’. I mentioned that you could decide whether to indicate paragraphs by vertical space (staccato) or no space and indents (traditional, more linear). Well, you can for main text, but not for displayed text, which was coded/rendered with a line space between each line of quoted verse.

Hyphenation. I left this on, and most of the time it worked. It broke capitalized words, though, and didn’t seem to mind breaking a very short word (the hero’s name is Syme, or Sy- | me as it was often rendered). The text is also short enough to read in about 3 or 4 sittings, so the lack of headlines wasn’t a bother (Stanza doesn’t bother you with an over-intrusive ‘where you are in the book’ indicator, though you can see a percentage read if you want to.)

So far, the Stanza implementation of Thursday was, as I said, reasonably happy. Better than a cramped pre-war Penguin. But as you can see, even in this text, anything that verges out of the pure linear interrupted mode starts causing problems. Enlarging the type, I forced the lines of poetry to turn over. The turns align with the start of the line rather than indenting – but while this is not as good as a hanging indent, it is, I suppose, better than having the start of the line indented with the continuation full out, which would really blur the distinction between prose and verse.

You will also have noticed from the screen-shots that the quote marks are straight throughout this text, which is a pity given the generally good appearance of the type.

To conclude, Michael Twyman’s schema gives us a useful checklist of the configurations that demand spatially sensitive, rather than arbitrarily reflowing, presentation. A list (and you can define poetry as being more list-like than continuous prose-like) needs its list-iness preserved. Headings are also a kind of list, in that they have text nested beneath them which, if hidden, renders the headings as a list of contents. A play or poem with line numbers is not continuous prose, but some kind of matrix, where the alignment of rows across columns must be preserved if the text area is resized. One could go on. But I would argue that being more analytical about the configuration that the individual elements that a text contains will lead to better designed eBooks through better design of the composition rules that render them.

Michael Twyman, ‘The graphic presentation of language’, Information Design Journal, Volume 3, Number 1, 1982, pp. 2–22

Diagramming remembrance

¶ Two interesting articles here and here.
With acknowledgement to Jessica Hagy