Abstract The 21 cm hyperfine transition of neutral hydrogen offers a promising probe of the large scale structure of the universe before and during the Epoch of Reionization, when the first ionizing sources formed. Bright radio emission from foreground sources remains the biggest obstacle to detecting the faint 21 cm signal. However, the expected smoothness of foreground power leaves a clean window in Fourier space where the EoR signal can potentially be seen over thermal noise. Though the boundary of this window is well-defined in principle, spectral structure in foreground sources, instrumental chromaticity, and choice of spectral weighting in analysis all affect how much foreground power spills over into the EoR window. In this paper, we run a suite of numerical simulations of wide-field visibility measurements, with a variety of diffuse foreground models and instrument configurations, and measure the extent of contaminated Fourier modes in the EoR window using a delay-transform approach to estimating power spectra. We also test these effects with a model of the HERA antenna beam generated from electromagnetic simulations, to take into account further chromatic effects in the real instrument. We find that foreground power spillover is dominated by the so-called “pitchfork effect”, in which diffuse foreground power is brightened near the horizon due to the shortening of baselines. As a result, the extent of contaminated modes in the EoR window is largely constant over time, except when the galaxy is near the pointing center.