Is this ->> an old operator or a typo/error?

It looks like a problem in the transcription process. There is a similar problem in DR 42, where the greater than sign is doubled: http://www.open-std.org/jtc1/sc22/wg14/docs/rr/dr_042.html

I learned C in 1992, and I'm 100% certain there was no such operator back then.

From the context, p->>x[5], we can deduce that it appears to do exactly the same thing as the more familiar arrow operator, ->. It is therefore likely to be a typo.

Alternatively, it could be an encoding issue in transcribing the code into HTML. If you look at the source to that page, you can see it's got a strange mixture of escape codes and literal < and > characters:

<TT><B>#include <stdlib.h><BR>

This does seem likely to have been a transcription error, but I think it would be useful to write out how a real C compiler would interpret this construct, anyway, just to make clear that it isn't a clever trick. The first thing it's important to know is this sentence, from C11 §6.5.4p4 (technically, N1570; this language is unchanged since C89, although the section number was probably different; emphasis mine):

If the input stream has been parsed into preprocessing tokens up to a given character, the next preprocessing token is the longest sequence of characters that could constitute a preprocessing token.

That means the six-character string " p->>x" must be tokenized as p -> > x, not p - >> x or p - > > x. (It doesn't actually matter in this case, it would be a syntax error either way, but this rule can be the difference between a program parsing as intended, and not; the standard gives the example x+++++y, which is interpreted as x++ ++ +y, not as x++ + ++y, even though only the latter is a well-formed expression.)

The next thing to know is simply that the right-hand argument of the -> operator must be an identifier, per the grammar rules for postfix-expression in §6.5.2. Obviously > isn't an identifier, so we have a definite syntax error.