A careful intercomparison is made of several sets of high S/N spectrsocopic observations made by different astronomers, using different telescopes and detectors, observing the same objects. Spectra of moderately weak (10-50 mÅ) Li lines, observed at S/N of one to several hundred using the CFHT Reticon, Lick Observatory Reticon, McDonald Observatory Digicon, and the Intermediate Dispersion Spectrograph plus CCD of the Isaac Newton Telescope agree within the claimed measurement accuracies of about 3-6 mÅ almost without exception. CFHT Reticon observations of Hyades dwarfs are compared with unpublished observations taken with a CCD at Lick Observatory; again, the agreement is very good. However, abundance determinations based on these measurements agree less well. The discrepancy arises from different temperatures used by different investigators for the same stars. The accuracy of various methods of temperature determination is examined, including B-V, R-I, and V-K colors, continuum scans, Hα line profile fitting, and temperatures derived from fine analyses. With careful work and good data, temperatures still limit the accuracy of some of the programs discussed.