The Bureau of Mines investigated the sources of error of a GLC-FID system in quantitative analysis of wide-boiling-range mixtures such as are encountered in simulated distillation of crude shale oil or petroleum. The sample-introduction system, column effects, and the FID itself were studied using normal alkanes as a diagnostic aid. The on-column inlet technique of sample introduction is compared with the flash-vaporization inlet technique. Column-adsorption effects and the use of tail-reducing materials are discussed. Problems with the FID relating to insulator and electrode contamination, electrode emissivity, sample capacity, and burner design are illustrated and discussed. Normal alkane FID response relative to n-heptane over the C7-C32 range is shown to be 1.00 +/- 0.03 for a parallel-plate FID when operating normally.