"The vibron signature of the H2 molecule persists to at least 316 GPa ..." ( Nature , Vol 416, p. 613)

Most of Next Wave's readers will be able to identify that sentence as referring to some sort of high-pressure molecular phenomenon. A somewhat smaller number may have a deeper understanding of the statement, even taken out of context as it is here. But the number that could confidently translate it into another language is smaller still.

Scientific translation involves an unusual set of aptitudes, including a broad grasp of scientific concepts, a highly specialized vocabulary in two or more languages, well-honed research skills, and the ability to reproduce challenging texts from a second language without sacrificing accuracy, clarity, or style in the first. Combine that with a deep sense of responsibility to the reader and the content, and an unfailing commitment to maintaining deadlines, and you'll have a fairly clear profile of the translators and Japanese language editorial staff at the Tokyo office of Nature magazine, where it has been my privilege to work for the past year. But rather than explaining in general terms what it is we do, I thought the clearest way to convey some of the demands of translating for a weekly science journal would be to walk through the process step by step, using the single sentence shown above as an example.

The translator receives the sentence on a Tuesday, as part of the first paragraph of a paper appearing in the following week's issue, which is published on Thursday. Her assignments usually involve chemistry and applied physics, but sometimes range into less familiar territory, such as information technology or conceptual papers on subjects like phase transition modeling. Fortunately, she knows her stuff, and an average 180-word abstract takes from 1 to 3 hours to complete.

The sentence at hand presents her with only one real challenge: a decision on how to render 'vibron' in Japanese. The word is effectively a nonentity. It's not listed in the Japanese Ministry of Education's Japanese Scientific Terms: Physics, nor in the authoritative Iwanami Dictionary of Physical Science. (There is an index reference for 'vibronic coupling' in the latter work, but no definitive listing for vibron. One suggestion is shinden, which combines the Chinese characters for vibration and electricity.) Based on experience from phoneticizing similar nonentities into Japanese, she renders it as biburon, under the fairly safe assumption that anyone in Japan likely to take notice will recognize the word as vibron written in phonetic script. She finishes the remainder of the translation, reads it again to check for typos and readability, and sends the data off to Nature to be edited for publication.

My job begins here. I participate on the translation checking effort at Nature , and my weekly assignment includes the paper in question. I am by no means a physicist, or even a scientist by training; I entered this line of work from the language side. After stints translating software instruction manuals and onscreen displays, I moved into more technical translations of software specifications, telecommunications glossaries, and intellectual property documents. For the two and a half years immediately prior to joining Nature , I worked mainly on a patient simulation software project at a major electronics manufacturer, a job that required fairly intensive study of physiology, pharmacology, and clinical medicine in both Japanese and English.

I start work on the week's translations on Sunday night, in preparation for a daylong discussion session with the three other translation copy editors on Monday. The editorial process involves reading English and Japanese texts side by side to confirm that both are saying the same thing. In the initial scan, my eye catches on 'signature' (which I have seen mistranslated in the past, but not here) and 'vibron,' which is an unfamiliar term for me, so I flag it in the Japanese.

I have no better luck finding vibron in the technical dictionaries, and decide to do a Web search. It turns out that the term was coined in the early 1980s to describe an algebraic model used in molecular spectroscopy. There are four viable options for rendering vibron in the Japanese syllabary: biburon, baiburon, viburon, and vaiburon, but the character used to approximate the English V sound is not generally recognized by the Japanese Ministry of Education, and we try to avoid using it in Nature when possible. I do a Google search on all four candidates, just in case, and get 116, 16, 1, and 0 hits, respectively. At first, it looks like biburon is the clear winner, but on reviewing those 116 hits, I mainly find pages linking to a newsletter for booklovers named Philobiblon, some herpetology pages describing certain species of boa constrictors and geckos, and, inevitably, a few adult Web sites. Scanning down to the bottom of the results page, I can find no clear reference to molecular vibrations. Baiburon does better, with 3 of its 16 hits clearly relating to the word vibron as a term from molecular spectrography. Viburon also makes a strong showing with its sole hit relating directly to vibrons.

I sign off on the checks around midnight on Sunday evening and e-mail the data to my office. The first set of files for the next week's issue has already started to arrive from the London office. Based on my inconclusive results, I decided to take my chances and change the phoneticization from biburon to baiburon. If I'm wrong, chances are good that a colleague will catch it before it goes to press. Or maybe a keen-eyed reader will spot it and earn the satisfaction of seeing a printed correction.

In this line of work there's room for doubt on some subtle point nearly every week. Should I have left it as biburon? Should I have gone with the Iwanami entry, shinden? Should I have ignored the usage guidelines and changed it to viburon? As a person responsible for communicating science faithfully across language and culture, you have to learn to live with uncertainty. But you never stop trying to get things right.