I suspect that your Fluke is not giving you accurate RMS conversion. There are a number of reasons that could be happening.
Scope the output as you have already done, and note the peak of the biggest clean sine wave. Then divide that by 1.414, the conversion factor for the peak to RMS value for a true sine wave. That number is your RMS voltage.
If your load was a resistor, the power would be the computed RMS squared divided by the resistance. I say resistance because speakers are NOT resistive loads. Their impedance has peaks and valleys.
100W into 8 ohms is P = V^2 / R; solving for P = 100, R = 8, we get V^2 = P*R = 100 * 8, so Vrms = 28.28, Vpk = 39.99; call it 40 volts peak.
140W into 8 ohms is 33.46Vrms, as you computed, and 47.32V peak.
Meters get funny about AC waveforms and frequencies, even "true rms" meters.