It would have been utterly unrealistic to expect one translator to work full-time for two or three years on a single project like this, so it was decided to split the job up. The correspondence runs to just under 1 million words, roughly two-thirds written in Dutch and the remainder in French. The most obvious division was three Dutch-English translators and two French-English.

The five translators selected from 59 entrants to an international competition were, in chronological letter order, Diane Webb, John Rudge and Lynne Richards for the letters written in Dutch, and Imogen Forster and Sue Dyson for those in French. Imogen Forster supplied the translation of French passages in the Dutch letters, while Lynne Richards did the same for the Dutch letters from the French period.

Another demand made of the translators was the rigorous avoidance of interpretation. If Van Gogh wrote ‘het’ or ‘il/le’ (‘it’), then ‘it’ it was, not ‘it [the money]’ or ‘the money’ if his meaning was not immediately clear. It would not have been apparent to the recipient of the letter either, so there was no earthly reason to remove any ambiguity for the modern reader. Clarifications, where necessary, will be found in the notes.

The translators supplied their work in batches. The first version was read by the translation editor, Michael Hoyle, (the first versions from the French were first read and commented on by Wouter van der Veen before being passed on to the translation editor) and returned to the translator with suggestions for changes. This second version, was then read by all the editors and emended into the semi-final version, often after long discussions of particularly problematic passages. When that stage had been completed, the translation editor went through all the letters in chronological order to produce the definitive version published here.

In the meantime, a 17,000-entry concordance was compiled to ensure consistency in the translation of specific words that Van Gogh used in differing senses. One such in Dutch was scharrelen, which could mean to potter about, rummage, try out, scramble, bustle, muddle along, struggle on, fiddle around with, mess about with or get along. In French there was nature, which was used variously for natural surroundings, landscape and reality, as was natuur, the Dutch equivalent in those days.

It was decided not to reproduce grammatical errors or spelling mistakes, and some were quietly corrected without drawing attention to them, such as ‘The Langlois bridge’ for Van Gogh’s ‘Pont de l’Anglais’. Nor were antiquated forms of proper names retained, so for example ‘De Gas’ becomes ‘Degas’.

We had to be particularly careful with technical terms and the names of colours. The same is true for descriptions of colours in which it is unclear whether Van Gogh is referring to truly different tones or to a simple mix. It makes a difference if one interprets ‘ocre jaune’ as two related colours or as a description of a particular shade of yellow – the difference being between a notional comma and hyphen. In some cases, the artwork under discussion can provide the answer, but this is sometimes complicated by the fact that pigments have discoloured or disappeared over time. Even when Van Gogh does not give a real title to a work, but only provides a quasi-narrative description of the subject or certain pictorial elements – and it thus remains unclear which picture he is actually talking about – this, too, had to be reflected in the translation. In such cases, the annotations can provide the necessary clarification.

It is a truism that shades of meaning get lost in translation, but needless to say we have tried to keep that to a minimum. A more troublesome area concerns a few specific features of Van Gogh’s use of language. His French was better than that of many foreigners, but it was not his native tongue.2626. See Van der Veen 2002 for an analysis of Van Gogh’s French. Some unusual phrasings are due to his Dutch. This occasionally led to idiosyncrasies in spelling, syntax and wording which could not possibly be reflected in the translation. Consequently, no attempt has been made to replicate inconsistencies, errors or un-French French, since the solutions would be arbitrary and result in unnatural English.

There are awkward passages – but they are awkward in the original Dutch or French. Then there are sentences that go on and on, with clause upon clause dangling from every conceivable purchase point, much like an Alexander Calder mobile. They, too, are reproduced as faithfully as possible.

We were also faced with problems that confront every letter translator, namely that the writer uses a variety of styles depending on the addressee, that his or her use of language changes over the years, and that every author has stylistic idiosyncrasies – to name but the most obvious. The variety of letter writers, styles and addressees demands an equal variety of registers in the translation. Gauguin wrote differently from Theo, John Russell differently from the postman Joseph Roulin.

This, then, is neither a literary nor an interpretative translation. Van Gogh wrote what he wrote. All we had to do was translate, not interpret or second-guess, not get in the way of the words. As a translator, though, one inevitably has to make choices out of several possible readings from time to time. In such cases we based our decisions on our knowledge of Van Gogh’s use of language.2727. See also Jansen and Luijten 2001.