For the case of spectrophotometers, we can easily calibrate the device(intensity wise) taking the light source inside the device as the reference, without any need of precise knowledge about its spectral properties. We can replace the light source with almost anything as long as it provides sufficient amount of light in the spectrum we want to work at and it's relatively stable, then recalibrate the spectrophotometer.
But when we build a spectrometer, a device that measures the spectrum of a light source, what we can use as an intensity reference other than a light source whose spectrum is already measured by an other calibrated spectrometer? How were spectrometers calibrated for intensity, lets say, when they were first made? Is there any light source whose intensity spectra is intrinsically known? Does such a light source need to be held in very specific conditions to be taken as an intensity reference(temperature, voltage, pressure, etc.)? Is there any simple reference that can be used to make a simple spectrometer which will be accurate to about +-5% in the visible spectrum?