Still reading Durrell (and now almost done with Balthazar), I ran across the word parquet used in the French sense of ‘prosecutor’s office’ and decided to look it up in the OED. Much to my surprise, it turns out to be a French diminutive of parc ‘park’; neither the OED nor the French dictionaries I’ve consulted explain the semantic transition. So of course I had to look up park, where I found a far more complicated etymology than I had expected (I’ve pruned some of the more remote twigs of information):
< Anglo-Norman and Old French, Middle French, French parc large enclosed area of land or woodland where one keeps and raises animals for the hunt (1160-74), enclosed place planted with fruit trees, orchard (c1220-78), mobile enclosure where one keeps livestock when they sleep in the fields, area thus enclosed (1269), large enclosed area of land or woodland maintained for the decoration of a castle or country house, or for pleasure or recreation, etc. (1337), fortified camp (end of the 15th cent. …), collection of vehicles which an army makes use of (1823 …), prob. < post-classical Latin parricus fence (8th cent. in Ripuar. Laws as parracus, but prob. earlier: see below), pen for animals (9th cent.), park, enclosure (12th cent. in a British source; from 13th cent. as parrocus), prob. < an unattested *parra pole, rod (cf. Spanish parra artificially supported vine, Catalan parra (type of) vine, Portuguese parra grapevine leaf; perh. ult. related to the base of Old French barre BAR n.1) + –icus -IC suffix. Cf. post-classical Latin parcus park, enclosure (freq. from 9th cent. in British sources), fence (12th cent. in a British source), pen for animals (freq. from 13th cent. in British sources), Old Occitan, Occitan pargue, parc, Italian parco …, Spanish parque …, Portuguese parque …, German Park (from early 17th cent. in travel writings, after English and French; 15th cent. in Middle High German in sense ‘compound, enclosure’; < French). Cf. PARC n.
Currency of post-classical Latin parricus earlier than the date of its first recorded attestation is suggested by the probable early West Germanic loan represented by PARROCK n., and also by the widespread currency of reflexes in Gallo-Romance dialects and in northern Italy. Most (although not all) recent commentators have regarded it as less likely that the word is a borrowing from Germanic into Latin (as frequently suggested in the past), on account of: (1) the initial p– (which is very rare in Germanic (see P n.), although it might be explained if the word were an earlier borrowing in Germanic from another language and not an inherited Indo-European word; this might also explain the fact that the word is recorded only in West Germanic); (2) the lack of evidence for a corresponding simplex in Germanic (although see PARROCK n. for discussion of PAR n.2 and PAR v.1); (3) the Latin suffix (although it is uncertain whether the vocalism of the last syllable in the continental West Germanic forms rules out the Germanic suffix -OCK suffix); (4) the possibility of Celtic cognates for the Latin word (although it is unclear what, if any, evidence there is to support the forms posited, e.g. Welsh parr enclosed place, Breton par plot of land); and (5) the possible connection with the base of Old French barre BAR n.1 There is thus no completely convincing argument for either a Latin or a Germanic origin. In favour of a Germanic origin, J. Corominas (Diccionario Crítico Etimológico de la Lengua Castellana (1985) s.v. parra) argues that Occitan parran enclosure, garden (a1168) is likely to be a borrowing from Germanic on formal grounds, and hence that the same is likely to apply for the other Romance words.
Welsh parc and Irish páirc are either < English or directly < French; Middle Breton, Breton park is < French.
Any other dictionary would replace almost all of that by the statement “There is no convincing argument for either a Latin or a Germanic origin,” but the OED lets you peer into the etymologist’s workshop and (if you are so inclined) decide for yourself where the probabilities lie.