Jisho

I'm sorry, first link is actually this one:

• https://www.nihongo-pro.com/kanji-pal/list/radical/92

There are two ways to write it, one with four strokes and one with five. You can see animations for both ways here https://kakijun.jp/page/kiba05200.html

Ohh! I see; thanks for the explanation and reference!

I'm sorry, also gave a wrong information about the first site. There were more appearances of 牙 (and its variations) than I had looked for, and so I just missed the next one (of 5 strokes). It's still a bit confusing, but I think I'm starting to get the right idea. Sometimes, the digital symbols do not match even though they look identical.

For the computer, this ⽛ and this 牙 aren't equal symbols. Sometimes, by getting kanji/radicals from different sources in the internet leads to this issue. Another problem, is that sometimes it is also very hard to find the correct variation (so, I cannot seem to get the symbol for the 5-strokes; but I'm pretty sure at least the one for 4-strokes is secured.

For some reason the Korean section of the wiktionary article shows the 5 stroke version, but if I copy and paste it, it turns into the 4 stroke version 牙. I guess it must be the font they use? https://en.wiktionary.org/wiki/%E7%89%99

If you're wondering about how a computer represents text, in most cases nowadays it's based on the Unicode standard. Some things about it are quite complicated, but for a basic understanding it's enough to know that one of the main things they do is assign every character a name and a "codepoint". Try it out here: https://unicodeplus.com

For example ⽛ is listed as "Kangxi Radical Fang (U+2F5B)", so the codepoint in Unicode would be 2F5B. And if you look up 牙 then it's "CJK Unified Ideograph-7259 (U+7259)", a completely different codepoint. All the codepoints are organized into different blocks (and further into planes) which you can think of simply as categories. That can help when trying to distinguish between two characters that look similar. Here for example you were asking about 牙 as a character on its own and not just as a radical, so the latter (codepoint 7259 from the "CJK ideograph" block) would be appropriate to use.

Note however that in this specific case, this isn't really the issue here at hand. Just because we have decided on a certain character with one Unicode codepoint, doesn't mean there aren't multiple ways of writing it. Just think about the Latin alphabet and how many ways there are to write an "A" or really every character of the alphabet. Unicode does not concern itself much with different ways of writing the same character (often actually there's some kind of grey zone, but not going to go into that here). In any case, as far as the computer and its character encoding is concerned, a 牙 is just a certain byte value, it doesn't care whether it's rendered to you with 4 or with 5 strokes.

So why does it look different sometimes? Well yes, it comes down to the font that is being used. Different fonts show characters in a different way. You probably already noticed this when looking at Simplified Chinese text in comparison to Japanese text. There are many examples of characters that are the same, but look different. Think 直 or 曜 or 骨 etc.
Even though they are encoded in the same way (for example 直 has the codepoint 76F4), when you use a Japanese font it will look very distinct from a Chinese font.

A website can (and should!) declare what language the text is written in inside its HTML markup by using a "lang" attribute on any tag. That way the browser knows which font to choose. That's why the Korean section in Wiktionary looks different from e.g. the Japanese section. They always declare lang="ko" for Korean language and lang="ja" for Japanese language. It's not so much that Wikipedia itself declares the font, it's just that they declare the linguistic context, and your browser then chooses an appropriate font.

Unfortunately not all websites, applications etc. do this the right way. Even some OSes screw this up. That's why some people argue that the Unicode standard should not have "unified" CJK characters by giving them the same codepoint even for different countries. If you want to find out more about this controversy just search "CJK unification" and there's plenty going back aeons. But whatever you think of it, it is here to stay and there's no real alternative to Unicode that doesn't break in a multitude of other ways. The end.

Post scriptum: Yes, there are also sometimes instances where computers don't use Unicode for representing text digitally. Really by 2024 the best thing to do is just switch to some encoding based on Unicode in most cases and if you have a special case then you already know that and are not looking for advice on that on Jisho.org forums.

Sorry about the delay. And thank you very much for the great insights in some of structural aspects of these issues. This is very helpful in understanding what can go wrong and why. Cheers!