The observed behavior of {eta} Car from 1860 to 1940 has not been considered in most recent accounts, nor has it been explained in any quantitative model. We have used modern digital processing techniques to examine Harvard objective-prism spectra made from 1892 to 1941. Relatively high excitation HeI{lambda}4471 and [FeIII]4658 emission, conspicuous today, were weak and perhaps absent throughout those years. Feast et al. (2001MNRAS.322..741F) noted this qualitative fact for other pre-1920 spectra, but we quantify it and extend it to a time only three years before Gaviola (1953ApJ...118..234G)'s first observations of the high-excitation features. Evidently the supply of helium-ionizing photons ({lambda}<504{AA}) grew rapidly between 1941 and 1944. The apparent scarcity of such far-UV radiation before 1944 is difficult to explain in models that employ a hot massive secondary star, because no feasible dense wind or obscuration by dust would have hidden the photoionization caused by the proposed companion during most of its orbital period. We also discuss the qualitative near-constancy of the spectrum from 1900 to 1940, and {eta} Car's photometric and spectroscopic transition between 1940 and 1953.