It's easy to make magic numbers on a dyno.
You always dyno in the early morning or late evening for best air quality. Late October, you say? All the better.
Measure the barometric pressure and enter it into the console at least 2-3 inches lower. If it's a beautiful day, then deduct 4-5 inches. Never tell the dyno you measured 30.2 inches when you can tell it 26.2 inches.
Watch EGTs and lean on the jetting and timing as far as you dare.
For the last pull of the day, you always remove the air turbine from the carb, 'just to show how much it restricts the air flow'. But here's the trick - you must be in the dyno room alone, so you can remove the inlet air temperature sensor from the turbine. Once it is removed, hang it down the back side of the motor, so it is close to one of the header collectors.
Now make the pull and when the recorded numbers are printing, distract the customer's attention long enough to print the corrected numbers sheet. The corrected sheet is what you want the customer to see, because it will be 'true Texas horsepower'.
Several years back, we had a customer and very good friend of the shop we did Comp Eliminator motors for, who decided he wanted to step up to Pro Stock. He's a rather impatient sort, but has the finances to back it up. He purchased two used 500" motors from two different shops in Texas. He had the sheets on both motors and brought both to us for a quick B.S. check, just to keep everyone honest. One motor had been built at Reher-Morrison and you could nearly lay our sheets atop theirs. We were <2 horsepower apart. The other engine was from a different shop (which will remain unnamed). We couldn't get within 20 horsepower of that one and beat ourselves up trying to sort the problem. Then we noticed the barometric pressure clearly indicated they had dyno'ed the motor in the eye of a hurricane, with air inlet temperatures approaching 200° F.
On another occasion, we dyno'ed a Super Stock motor that was still warm from Lingenfelter's dyno. Our numbers were identical to what John had recorded.
It's sad to say some shops are willing to toss ethics out the window to make themselves look better than they really are. A SuperFlow printout will show a theoretical number that is calculated to show an approximate parasitic horsepower loss figure. Obviously, you cannot measure something that wasn't there to be measured, but some dyno operators will add the parasitic loss number back to what was measured to pump up the number.
When it comes to a chassis dyno, you are measuring rear wheel torque/horsepower. With that number, there is no way to accurately determine horsepower at the crank flange. You can guesstimate the number, but that is as good as it gets. Unless you know the exact percentage of coupling loss for your particular torque convertor or the percentage of clutch slippage, the parasitic loss of the transmission and the parasitic loss of the differential, all you have is a rear wheel number.
And we're not addressing the issue of calibrating the absorption unit in the test cell itself. It's a bit like that torque wrench in your toolbox - when was it last calibrated?
A dyno is a tool to measure power and torque, so engine output numbers can be re-evaluated after making engine changes. Did it get better or worse? At what RPM range the change appear? Hang a heavy balancer on the motor and the dyno will make you smile. Take that motor to the racetrack and the motor will accelerate like a slug.
The only numbers that are real are the numbers you recorded on a given dyno with a given motor on a given day. Anything else is guess-work.