Do I need to scale an Inverse Matrix by it coefficient

IMK

Hello,

http://www.summitinstruments.com/kno.../pdf/TN413.pdf

So I have had a go at implementing the math on page 1 of the attached document and it
seems to work. I say seems to work because the way that I gather the accelerometers calibration values (offset and scale) may not be perfect hence I could be introducing my own small errors. Also I only have a limited set of test data ( R at different static orientations) to test with.

My question is:

Do I need to scale the matrix T or B by the coefficient of T to get the correct/best answer?

I ask as when I do scale B by its coefficient the error size is smaller.

I measure the error by 1 - |R| = error size.

Many thanks in advance