OK so I assume your two spectra are scaled relative to the continuum over some common wavelength range? From the AAVSO data however, the V band total flux has reduced by ~0.4 mag between the two dates (I assumed values of 8.3 and 8.7) which means the amount of flux in the emission lines in the later relative flux calibrated spectrum is exaggerated. To calibrate in absolute flux I used a tool in ISIS (“Tools”, “Spectra 3” , “Flux density conversion” ) which integrates the flux in the spectrum over the standard (Bessel) V filter passband and uses the photometric V magnitude (from AAVSO) to rescale the spectrum in physical flux values. (Using the relationship between V mag and physical flux established for the standard star Vega). There are some approximations/assumptions in this method I believe and I think David Boyd has published a more rigorous method somewhere where he uses his own photometric brightness measurements but I think it should be good enough to show the trends.