If you were to measure the spectrum of an object while it's exactly at the zenith, how should you orient the slit of the spectrograph to minimise atmospheric dispersion? Is North-South or East-West better? Or does it not make a difference either way?
If you are at zenith, it doesn't matter - the light coming through the atmosphere is not differentially refracted.
If you are away from zenith, then the position of the star in the sky depends on wavelength due to differential refraction. This differential refraction is along a line from the star to the zenith - at the so-called parallactic angle to a N-S line. If you are doing spectrophotometry or spectroscopy over a large wavelength range, it's usual to rotate the slit so it is parallel to this line . If you don't, then the danger is that some wavelengths of light won't fall in the slit at all. The size of the effect increases with angle from the zenith.