"The C++ programming language", third edition, starts its appendices on page 791, its index on page 869, and ends on page 910.
Both use about the same line spacing; the C++ book is about 10% wider and its lines are about 15% longer. It also has fewer comparisons with older omplementations such as MacLisp and Lisp Machine Lisp, and maybe a tiny bit less verbose.
I do not think either is THE standard, but they should be representative. So, there is nowhere a factor of ten.
Common Lisp is large partly because it has everything you would ever need (print the integer 4 as 'four', as 'fourth', as 'IV' or as 'iiii'? You got it. Pluralizing strings? You got it). The loop macro (a for loop on steroids, with all its positive and negative effects)? 38 pages. Pretty printing? 22 pages.
The spec covers not just the language itself, but the whole standard library.
> The R5RS Scheme specification is 50 pages.
And with those 50 pages, you can not do anything useful out of the box. Yeah, you can solve SICP excercises, but thats essentially it. All useful features are implementation specific, and the Scheme ecosystem is more fragmented than the Balkan. Scheme code is essentially non-portable amongst the 1001 existing implementation. You dont write "Scheme" directly. You write code either for "Guile" or for "Chicken" or "Bigloo". PLT doesnt even call their implementation Scheme any more, they call it Racket. You dont write Scheme, you write Racket.
This leads to the question of why a standard was nescessary at all, if youre not standardizing a critical amount of features to enable portable code.
If this[0] is a valid copy of the spec and my calculation is right [1] then it's 1360 pages. Seems a lot, but I have no idea what the length of ANSI standards is.
[0] http://quimby.gnus.org/circus/cl/ansi-cl/ a draft but i think close to the final thing
[1] for i in `ls *.ps`; do gs -q -dNOPAUSE -dBATCH -sDEVICE=bbox $i 2>&1 | grep -c HiResBoundingBox; done | awk '{ sum += $1 } END { print sum }' , and each ps page is actually two