astronomers + free breathalyser = ?

When recently staying at the Coachman Motel in Parkes for a radio school, we were1 incredibly delighted to stumble across a free breathalyser in the adjoining Coachman bar. I didn't think too much of it at the time, aside from being awed by its excellent price. But later as we left the Leagues Club, I saw the opportunity for some experimentation. My table (although reluctantly at first) had been on the pink moscato basically since the dinner started, and so I kept my last mouthful with me as we headed back to the Coachman. My trusty co-experimenter Izaak2 and I headed for the breathalyser to blow some calibrated scales.

The plan? Take as many (free!) measurements as we could stay awake for in 5 minute intervals, and then analyse and plot the data later. I drank nothing more, in order for the data to trace only the decay in BAC over time. We started our measurements at roughly 10:45pm, and made it till just after midnight. The party sure was happening at the breathalyser that night - although I'm pretty sure the only ones that found it so amusing were us (radio astronomers) and the mining engineering students from UQ. Still, we kept coming back for more! It seemed to become a trophy if you scored a "DANGER!!!" from the breathalyser, since it meant your reading had greatly exceeded the upper reported limit of 0.1 and was thus beyond the scale. I only got this once, which was the reading taken just after I swallowed the carefully transported mouthful of moscato.

Of course, you are supposed to wait 10 minutes before taking a reading, so we knew this was always going to be a shaky data point. Nonetheless, we took it, and then proceeded to get the rest of the data at proper intervals that more realistically reflected how my BAC was behaving. For no particular reason, I expected the decay to be exponential and so was surprised to later find that it wasn't quite trending that way. In total we took 13 measurements, but the true value of the first reading is somewhat uncertain and probably not particularly meaningful anyway. To test this, I made plots and fits both including this point and without it, and you can see for yourself that there isn't a considerable difference. I used my camera to record each BAC reading, which had the added benefit of keeping time pretty well.

The plots are shown below:

The first plot includes the first data point in the fits while the second point excludes it for the sake of investigating its effect on the slope fitted to the decay. I used Matlab to estimate a simple polynomial fit to the line with the raw data (black lines), and then interpolated with the spline setting to smooth the fit (red lines), although it didn't have too much effect as you can see. But it did make the polyfit() function a little bit happier. This gave me the coefficients of the straight line approximating these points, and so we can then put an estimate of at what time the BAC would reach a certain level (equations plotted on the figures). It's probably a bit longer than predicted, since we don't know if the trend continues to be roughly linear at lower BACs.

If you're scientifically minded, you're probably wondering about error bars right about now. Yes - I wondered about them too. You can see from the plots that the BAC measurements are quite unstable, and fluctuate anywhere up to 0.01 from the seeming line of decay. I use the error of the first point to give a little bit of insight into the errors: the first point included suggests a minimum time till 0.00 of 3.68 hours, while without it suggests a maximum time of 3.91 hours. That is a difference of (3.91-3.68)*60 minutes, or 13.8 minutes which rounds to about 15 minutes. One goal of the plot is to estimate the time till a particular BAC, and so an error of 15 minutes is useful to take into account though it isn't too serious.

More significant here is if a predicted BAC of 0.00 could exhibit the same fluctuations of up to 0.01, which would suggest that we would want to extrapolate to a BAC of -0.01 to ensure that it had actually reached 0.00 (and similarly for 0.05). How much time difference does this add? A minimum of 4.07 hours (with the first point) and a maximum of 4.33 hours (without it). More important is the difference between the maximum time for a BAC of -0.01 and the minimum time for a BAC of 0.00, which is (4.33-3.68)*60 minutes, or 39 minutes. This would say that if you absolutely wanted to be sure that your BAC was not 0.01 higher than the measured reading (due to random fluctuations), then you'd need to wait about 40 minutes. I did the same calculation for 0.04 and the difference is about 30 minutes (due to an increasing divergence between the spline-fit and the raw-fit and the slope difference).

Of course, it might not be the machine's fault that the readings are fluctuating. It could also be that alcohol is not processed completely linearly in the human body, and that the variability we saw was actually quite real. This seems possible, and is something I'm hoping to investigate further in future! We had many ideas for expanding this preliminary study, and you'd be surprised how many volunteers put their hands up, which means that I don't have to be the guinea pig next time :p It would be cool to investigate the effects of different types of alcohol, or how your BAC measures when you continue drinking throughout the measurement process, or the effect of eating during the measurements. Will definitely be following up on this when I get a chance!

In conclusion, much fun can be had with a free breathalyser, and I strongly encourage all pubs to install one. Not only does it allow endless hours of scientific enjoyment, it also serves to educate people on the correlation between how much they drink, how they feel as a result, and what the actual alcohol concentration in their system reads as. Given the reasonably high profile of drink driving, it seems that these breathalyser machines should be provided free at all venues that serve alcohol. I hope to see more of these free breath testing machines in future!

update 02/10/2011
Immensely disappointed so far with the Digitech "Fuel Cell" breath tester we bought from Jaycar today. Although I was initially impressed with its apparent 0.001 accuracy, it showed barely any reading when I drank a whole glass (about 200 ml) of Baileys, except for maybe about 0.01, which is significantly lower than what it should have been. A couple of guinea pigs, who had drunk enough vodka to be in the 0.15 range, only registered 0.07 and 0.03 respectively. It seems to be considerably wrong in its calibration or it might just be shit in general. It's a shame, because I was hoping more experiments could be carried out with a not-too-expensive breath tester. Will have to take it back tomorrow :( Hopefully more experiments to come soon!

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License