Where should the type information be 118

Well, I might be misremembering, and it might have been 1980 :-)

Regrettably, those references aren't precise enough to be sure which concept they are referring to. The Multics one could have been using them as synonyms or as terms for two very different mechanisms.

That is clearly using them for descriptors. 1977.

Well, the first paragraph on page 10 reads to me as if it is talking about Iliffe vectors and not descriptors! I can't see that it is even ambiguous. 1971.

My belief is that the term was originally used as a near-synonym for Iliffe vectors up until (say) 1970 in the USA and 1980 in the UK. It then got attached to the descriptor concept; can you confirm that YOU used the term in the 1960s or early 1970s to mean descriptors?

I do know that there was and is very considerable ignorance in the USA about the very existence of descriptors - this was very visible in the late 1960s and early 1970s in some of the opposition to Algol 68 (despite PL-I), and again in the 1980s with some of the opposition to Fortran 90 and the way that multi-dimensional arrays were introduced into C++. Even within the past decade, I have educated major vendors' compiler teams about why and how they can be implemented efficiently.

In the UK, even computer scientists tended to know about them, but were often very prejudiced against them (which accounted for the C++ and RISC integer multiply debacles). It is unclear how much LISP and Pascal were to blame for this, but it was very clear that there was a major gulf between the "numeric" people who understood matrices and the computer scientists who didn't.

Regards, Nick Maclaren.