I remember a conversation with an IBM engineer back in the 1980s, who implied there was an internal fight over this between IBM's engineering and marketing departments.
The engineers wanted the PC to have the same keyboard as the popular 3270 mainframe terminal, for easy migration of users and software in a business environment, and in fact IBM did produce the "IBM 3270 personal computer" with this keyboard.
However the marketing guys wanted something smaller and cheaper than the 122-key 3270 design with independent cursor keys and numeric keypad, plus 24 function keys above the main keyboard and a further block of 10 special-purpose "function keys" to the left of it. The original 3270 keyboards were heavy as well as big, since they had sheet metal cases rather than plastic.
The original IBM PC keyboard repurposed the 10 left-hand keys as general purpose function keys, deleted the original 24 function keys, and compressed the layout of the right hand side to save space. The PCjr keyboard went even further and reduced the total number of keys down to 62.
These PC keyboards were not popular with users who were accustomed to mainframe terminals, and eventually the 101 or 102 key "enhanced keyboard" design emerged as a compromise, and remained as basis for the current PC "standard keyboard."
See http://www.quadibloc.com/comp/kyb03.htm for layouts of the various keyboards referred to.
I suppose the idea of mapping the cursor keys onto part of the main keyboard with a separate "function key" to toggle that behaviour simply didn't occur to anyone at IBM at the time, coming from a "big keyboard" background. Even now, laptop users who want to use keyboard-intensive applications usually buy an add-on keypad.
+
shape, as opposed to inverted-T
, is not at all comfortable.