But this article is excellent. Even if it also contains opinions of the author about "aberrant French practices", some of them are still prevalent such as the persistant use of an "extra" spacing before colon, semi-colon, exclamation and question marks, and within guillemets: - the practice is still used by almsot all French publishers and in newspapers, even if most English publishers ignore it; in modern French applications for computers or the web - it is limited by the support of multiple space characters, so that the oly choice is between breakable an non-breaking space, but the reality is that French publishers still want to use (non-breaking) "narrow spaces" ("fine" in French, about 1/6 to 1.5 em) with these punctuations. - the author recognize that these practices were also within recommandations of English typographs, but English publishers today have forgotten this rule, as well as users of typewriters, and most modern texts on computers and on the web ignore (or don't want) these thin spaces (when there's no support for them), while French users prefer using (non-breaking) normal en-spaces even if it's too large, to using none.
However modern typesetters using computers are muchless constrained than publishers that were using metal types : the glyphs of metal types could not be stretched like what can be done easily today, to adjust lines forjustification (and limit the formation of infamous vertical "rivers" across lines of text). So what can we conclude: all spaces are variable when typesetting, but not the same way depending on their context. But this is something that belongs to typesetting, not to the initial "plain text" to render as it needs to be understood when reading. These questions are not just about "esthetic", but about preserving the average blackness of lines to guide the eye for easier and faster reading, and to make sure that important punctuation will be easily distinguished (because they guide the "rythm" with which the text should be clearly read by speech (imagine you're reading the text to a public with clear voice, for better understanding: this is not an evident practice, good readers are rare that can translate to their auditory the substance of the text with emotion and strength as it could have been intended by the author, better exhibiting his choice of words). But if we have a plain-text to typeset, an automated typesetting program that will convert these spaces and balance them to respect these helpful eye-catching rules need to be hinted. If the plain-text only contains single spaces, the typesetting program will be fooled. The minimum set of spaces to insert in text must first distinguish breakabke and non-breaking spaces. Then it must be able to look at contexts before and after the space to se ehow to adjust them, **if it is needed** (modern typesetting programs should probably use moderate glyph stretching as it helps preserving balanced interworrd spaces, and it is probably better then trying to use multiple widths between spaces in the same line). The author of the article suggests that the minimum interword space is 1/3 em. This is probably true for English because its words are relative shorter than in French or German (larger minium spaces will more easily create "rivers"). In French the minimum space is preferably 1/2 em and in German (or languages that use complex compounds or agglutination of morphemes) it may even be up to 2/3 em. This has a consequence : spaces to separate sentences will unlikely need to be as large as in English (that's probably why French does not use wider spaces in those positions between sentences). And another consideration is also important here, notably the average length of full sentences (this is basically a property of the language, but also of author's own style). English sentences (by classic authors) are generally shorter, as English authors do not like sentences with multiple verbs or long enumerations and may adjectives or qualifiers (their style is often more "directive"). But modern technical or legal texts (includinc contracts, licences) are as complex and long as in French or many European languages (if not longer, because "short" English is often more ambiguous and could lead to dangerous misinterpretations!), so these type of texts tend to have only once (long) sentence per paragraph. (In that case the size of spaces after period does not matter because it oocurs before a paragraph break). And long enumerations within sentences (common in French), are preferably using bulleted lists (possibly in multiple columns) in English, within their own block separated from the introduction sentence. Typical English sentences for example will more rarely contain more than 2 commas, and not more than 2 verbs. If enumerations are used, they are preferably at end of sentences, in sentences containing no more than 1 verb (in German it would be exactly the reverse, with a more likely passive form to better place that verb if needed where English would use the active form). In summary, the typesetting rules are not as strict as one could think if you look only at the plain-text alone. It is adapted to the actual content of text, and to its style and the intrindic nature of the language, its phonology, and the way it is spoken for better understanding (people tend to read texts the way they would like to hear them, they mentally "vocalize" them because it helps understanding, they don't just read morphemes and so like to see the correct punctuation and typesetting should also reflect how they think and understand the important and structuring items in sentences). 2013/9/14 Michael Everson <ever...@evertype.com> > On 14 Sep 2013, at 19:11, Jim Allan <jallan...@rogers.com> wrote: > > > See http://www.heracliteanriver.com/?p=324 which claims with numerous > examples that Michael Everson is totally wrong. > > It's what I was taught. > > Michael Everson * http://www.evertype.com/ > > > >