Recalling my years in engineering school a million years ago, even with "accurate" equipment and testing, test results were typically nothing more than an average of values. There wasn't any "one true answer" to anything. The more data I had, the more sure I was of my results (i.e., smaller variances) and lots of data helped to spot trends and discard bad data. If the OP cares to repeat this experiment ~3 billion times, I think we'll be able to come to a reasonable conclusion w/o the power meter.
