I began with the root R-W-F in the Hans Wehr Arabic-English dictionary. It's not specifically a J-A dictionary, but it's where many people start. (It's also not a medieval dictionary, and while that's still where many people start even with medieval texts, I was translating a few lines from something modern.)
Unhelpful. And now I'm going to have the song lyric !الروف، الروف، الروف مشتعل stuck in my head, in Arabic, for the rest of the day.
Next I turned to Joshua Blau's dictionary of medieval Judaeo-Arabic texts. It is a remarkable addition to the research tools in my field, but it is also seriously incomplete, as it is based on a corpus of the texts that Joshua Blau has edited.
R-W-F reminded me that the middle radical can sometimes change to ', pointed me to R-'-F, and remarked to see section 97 of Blau's grammar for a discussion.
R-'-F reminded me that the middle radical was originally W, pointed me to R-W-F, and remarked to see section 97 of Blau's grammar for a discussion.
Finally, under R-'-F in Hans Wehr, a possibility:
But, no, the orthography is still not quite right. There is sometimes slippage with one weak letter dropping out of a doubled pair, but that seemed unlikely in this case, as one of the Ws is the seat of a hamza on top of the fact that apparently, this word is unattested in an adjectival form ending in -Y, as the orthography in the document in question has it.
A small eternity later, I realized that the word, appearing on a salary table from a Jewish community in modern India, is "rupee."