If the motor is running uncouple to the driven machine, I would be a little concerned with the 9 rpm slip on an uncoupled motor, that is if the spectrum's cursor is displaying the correct 1x rpm.
Sometimes, if the "Locate Peak" function is used to locate a frequency, it can give a, somewhat, wrong frequency, depending on the resolution and the amount of energy "riding" next to the frequency in question, in this case, 1x rpm. The use of the "toggle cursor" function to help align the fundamental and the harmonics of the fundamental, IMO, gives, most of the time, a closer true rpm. But then again, the resolution has an effect on this also.
I am referring to the CSI (Emerson) software. No problem with the software, IMO, just possibly the resolution used.
Have you strobe or confirmed the speed in the data you are posting?
Only my opinion and I could be totally wrong.
Thanks and Have a Great Day,