Article 1442 of comp.sys.handhelds:
From: dan@Apple.COM (Dan Allen)
Subject: HP-48 Details
Date: 13 Mar 90 03:05:26 GMT
Organization: Apple Computer Inc, Cupertino, CA

A few miscellaneous notes...

UBASE - I believe UBASE has a bug in it.  It always rounds fractional
unit powers to the nearest integer.  For most "normal" units it will not
bite you, but it did me when dealing with various geometricized units
used in general relativity...  All of the arithmetic operations seem to
preserve the powers of units fine.  It is only when using the UBASE
command that this will occur.  Example: 1_m^1.5 UBASE  --->>  1_m, which
is wrong.

ABORT - There is no ABORT command in the HP-48SX, as there was in the
HP-28.  However, KILL works the same.

VAR - The HP-48SX supports the statistical variance function, but it is
nowhere documented in the manual, but like the HP-28, it is there and
works fine.

BYTES - Does a good job as a hash function.

RND and TRNC - These are about the only functions which operate on every
element of a matrix (array); most other functions (like SIN, LN, etc.)
will only return an error if given a matrix.

TAYLR - Improved from 28S, as it returns denominators with their
factorial arguments unexpanded, i.e., 1/3!+1/5! instead of 1/6 + 1/120.
EVAL or ->NUM will coelesce the terms if needed.

0^0 = 1 - This is a nice improvement.  Most HP calculators have always
made 0^0 an error, but Don Knuth and W. Kahan have both persuasively
argued that 0^0 should be 1, which it is.  Nice!  (It makes number
theory stuff work right.)

Dan Allen
Apple Computer