I'm trying to do a lab writeup, and I'm going to include the uncertainty values as part of the data table. I was measuring distances with a meter stick with smallest divisions of 1mm, and estimated the error to be .05cm (.5mm - half the smallest division, as our TA told us was a good rule of thumb). However, having the data in (cm) was not useful, so I simply converted to (m), and the error along with it - the error of .05(cm) became an error of .0005(m). The TA, checking our data before we left the lab, told me that the conversion was wrong - that error can't get smaller just because you up-converted the data. I thought this was crazy because %error is still the same. She even pointed out the error propagation tables, showing me that the error for Z = a(B) ± C is dZ = a(dB), which seems to prove my point, not hers. Who is right?