Bishop, S. C., Johnson, G., Smith, L., D. D. Fiorentino, Garcia, T., Garcia, R., Breyer, C., & Loomis, W. D. (2009). Journal of Analytical Toxicology, 33, 521-524.

Utilization of a manual sampling function as an alternative to the automatic sampling function in the Alco-Sensor IV Black Dot Model has been recognized by the manufacturer to potentially underestimate an individual’s true breath alcohol content (BrAC). A controlled human subject study was conducted to analyze the possible breath-sampling differences between the standard automatic technique and three manual techniques. Subjects were dosed with vodka and orange juice and then tested during the descending limb of their BrAC curve. Differences between the automatic and the manual techniques were found to be statistically significant with the three manual techniques underestimating the BrAC. The average maximum difference between the automatic BrAC level, as compared to the lowest manual level in each data set, was 27.9% (median 27.7%) with underestimations from 20.8% to 40.0%. In no instance did any of the manual techniques produce higher BrACs than the automatic technique.