[[Appendix:Proto-Slavic/zǫbъ|zǫbъ]], [[Appendix:Proto-Balto-Slavic/źambas|źambas]]

I agree, but the difference between źombos and źambas is really trivial. o and a were the same phoneme in Balto-Slavic, so they are two letters for the same reconstructed sound. The only case I can think of where the difference between them would be significant is for a sound change that affected o and a differently. There is in fact such a change, Winter's law, which lengthens the vowels to ō and ā (which don't merge), showing us the distinction. But the only conclusion you can draw from that is that Winter's law occurred before the vowels merged. As far as we can tell, the merger happened before Balto-Slavic split apart, because all Balto-Slavic languages show only a single reflex of both original vowels (a in Baltic, o in Slavic but with evidence of an earlier a > o change in Middle Common Slavic), and there are no post-Balto-Slavic sound changes that require or show evidence for a distinction between a and o. The conclusion then is that they must have been indistinct in Proto-Balto-Slavic itself, and that any difference between them is purely notational.

And I don't dispute your idea here. It's just that, if this is your idea, then it should be sourced to you, not to Derksen, who had *źombos. Even if the difference is purely notational, we can't change what the source says. It would be like respelling a quote in American spelling despite the fact that it comes from a British source and uses British spelling.

Maybe this is a general question, even outside of reconstructed forms. In general, it seems Wiktionary practice is to respect the original form, no matter what. I see, for instance, that Serbian words are given both in Cyrilic and Latin scripts -- even though they are clearly notational variants of each other, Wiktionary doesn't just pick one of them and avoids the other (since both are in use). I see also pages for languages in the original script (say, Gothic, or Japanese), as well as pages with the same words in Romanized form.

So I'd imagine that if there is variation (*źombos vs. *źambas in different reconstructions) or of opionion (he has *źombos but my knowledge of the literature / my personal opinion leads me to think that *źambas is better) could be treated similarly. And only *źombos would be sourced to Dirksen, while *źambas would be, say, to you.

My point here is to avoid confusion. Precisely because there apparently already is confusion in the literature (different theories, different notations, different reconstructions, etc.), a new Wiktionary page about a reconstructed protoform should make an effort to justify its existence with this particular spelling / notation (rather than a variant); or else, we're adding an extra bit of confusion to what is already out there. Don't you think?

It's not my idea, I just don't have sources to back it up at the moment. It comes from things I have read about Balto-Slavic reconstruction.

But I don't think this should be equated with misquoting. If one source writes a physical equation as E = h ν, another as E = h f, and a third as E = ħ ω, then those are not three different equations that deserve separate treatment, because they express the same mathematical idea, just with different notation. In the same way, źombos and źambas are the same thing, just written differently. The crucial difference between this and the spelling of a real language like Serbo-Croatian is that one is actually used as a means of communication, whereas reconstructed languages are written in an abstract linguistic representation that just happens to resemble a written word (for convenience). But you are probably familiar with the custom of writing h1, h2 and h3 for three Indo-European phonemes, which shows it more clearly. As long as linguists agree on a representation, any will do. If a linguist had decided at one point that what we write as ź is to be spelled as Z or 2 or something else, then that would not have changed the reconstruction; only the way of representing it abstractly. You really can't treat it the same as a real written language, because the spelling of a real language is based only on convention, whereas linguists routinely make up their own conventions when it suits them, and what counts is the idea that underlies the letters, not the letters themselves.

I do agree that we should quote the source accurately, but I don't think that should imply that *źombos should appear in etymology sections alongside *źambas. Instead, the form *źombos should be placed with the quotation in the reference section. Reconstruction:Proto-Balto-Slavic/źombos should redirect to Reconstruction:Proto-Balto-Slavic/źambas if there is any chance that people will come here looking for *źombos.

Something like a redirect is OK by me, in principle. But, in Wiktionary, aren't spelling variants usually treated with a specific page that has a specific template like {{alternative spelling of}} plus a redirect to the preferred spelling?

And, even though it is true that reconstructions are more like formulas than like actual words, still they are words, which is why they are here at Wiktionary (I note formulas do not have pages here, which again is OK by me). And to say that *źombos is "the same thing" as *źambas in a way that is different from the way in which, say, plow and plough, or 𐍅𐌰𐌹𐍂 (wair) and wair, or zob and зоб, are "the same thing", is probably metaphysically correct, but in actual practice not very much. In all those cases, the reason for having both forms is that "both are used" (likewise both *źombos and *źambas are used, if I understand you correctly), and that an interested user could look for one of the forms, not knowing that the other also existed, and s/he should be able to find it (likewise for *źombos and *źambas). (Likewise, if Wiktionary listed formulas as words, it should indeed treat E = hν and E = hf as spelling variants, with independent pages, and E = ħω as something slightly different, a "rescaling" or "change of units", since it includes the 2π constant to allow the use of angular variables rather than non-angular ones, which is not the hace for hν and hf, two pure notational variants; but still with a separate page plus a link to the "default" formulation.)

I wonder if I should raise the issue somewhere else. There may be a need for more precision in the formatting of pages for reconstructed protoforms; apparently there aren't guidelines to it the way there are for normal lexical pages in various languages. --Pereru (talk) 23:00, 27 July 2013 (UTC)

Yes, both are used, and that's why the redirect can be created. "Alternative form" is ambiguous when it comes to reconstructions. In some cases, there may be two forms that can both be reconstructed and both were demonstrably used by speakers at the time; this would be an alternative form in the same sense that we use it for regular languages. But it could also just mean an alternative reconstruction from a different source, and that speakers really used one or the other but we can't tell which. That isn't really an alternative form though, more like an alternative hypothesis/reconstruction. For the entries I created, I have used "alternative form" entries only in the first sense, and created redirects for the second type. That was the common practice for Indo-European entries before I came to Wiktionary, so I adopted it as well. Several of our PIE entries list alternative reconstructions either on the headword line or in a usage note.

Also, here's a more general question. I see that alternate respellings of reconstructions (or even totally different reconstructed forms) already are in use in etymology sections of individual words, since I have come across some myself. So: should some kind of concerted effort be made to make everything consistent, i.e. should someone go looking for the "wrong" spellings of some reconstructed protoform and replace them with the "right" spellings? And what do we do if a source is cited, but it has the "wrong" spelling? (Maybe put the "right" spelling next to the "wrong" one in parenthesis? Or maybe leave the "wrong" spelling but link it to the "right" one?)

For several reconstructed languages we have a standard notation that is based on common practice (but never all; there will always be some that use another notation). That notation is described on the "about" pages, like WT:AGEM or WT:ASLA. There have been a few occasions where our standard notation has changed. The most notable is probably the use of ogoneks for Proto-Germanic. If etymologies use another spelling, I usually change them to reflect our standards. So far, I haven't had any that were sourced, but if they are, we should still use our own notation, and use the source's own notation when quoting it. Our etymology sections never directly quote sources, they are always paraphrasings, so converting the notation can just be part of that. I suppose it's assumed as understood that varying notations don't imply different reconstructions.