137Cs irradiators are used for many different applications including food processing, radiobiological experiments, and calibrations of radiation detectors at standards labs. Air-kerma modulation techniques including varying source size, adding attenuation, or varying source-to-detector distance, combined with the geometry of the collimation in the irradiator, may change the energy spectrum of the irradiator. This is unaccounted for during beam characterization, which is completed using a spherical ionization chamber with a minimal energy dependence in the 662 keV 137Cs gamma-ray emission region. This work incorporated a high-purity germanium (HPGe) detector collimated using a lead aperture to measure the energy spectrum of a 5 Ci source in a Hopewell Designs G-10 dual source irradiator with 4.08 cm and 5.54 cm of attenuating material. The measured pulse-height distributions (PHDs) were compared with simulations in MCNP6 to provide validation of the geometry of the irradiator model. PHD measurements were corrected for detector response using MCNP and the true spectrum was obtained using Gold's deconvolution method. Due to the fluence limitations and collimation of the HPGe detector, measurements represented the primary photon energy spectrum. The impact of scattered photons on the energy spectrum was investigated using MCNP6 simulations. In addition, simulations were run to determine the effects of attenuation and source-to-detector distance (SDD) on the energy spectrum. Comparisons of spectra were made in terms of mean energy. Simulated and measured PHDs showed good agreement when comparing spectral features including FWHM of the 662 keV photopeak, mean energy, and the total area under the curve. After deconvolution of the PHDs, the primary energy spectrum mimicked an encapsulated 137Cs source with emission of only the 662 keV photopeak. For 5.54 cm of attenuation, simulations showed that if scattered photons were taken into account, the average energy decreased from 662 keV to 656 keV, which was a 2% difference. Simulations showed that as attenuation increased from 0 cm to 5.54 cm, the mean energy of the spectrum increased 9% from 605 keV to 656 keV. The SDD did not affect the mean energy. This work used spectrum measurements to validate a 137Cs irradiator model in MCNP6, which represents a method that is more sensitive to geometry errors than conventional PDD and profile comparisons. Using the validated model, simulations were run to improve the understanding of the output of the Hopewell G-10 137Cs irradiator.