I wondered this also. I'm guessing at that time, they perhaps didn't have MUL or DIV instructions, or even the concept of SHL or ROL which would have allowed a test for "divisible by 2".
So I'm guessing they started with a "factor", example 2, added it together repeatedly until it equalled or was greater than the target, remembering how many additions were made. If equal to the target, repeat the addition half the number of times, and replace that with the old target, and output 2 as a factor. If greater than the target, try the next "factor" 3,5,7 etc etc.
Repeat until target = 0
It's surprising how relatively recent native MUL and DIV operations are in processors.