What is the legal limit of alcohol in the sweat of a person, so that it is deemed illegal for him to drive? Im more concerned about the legal limit in the UK, but any other country is also appreciated.
There is a certain amount of variation in the wording of basic drunk driving laws. For instance, in Washington state, RCW 46.61.502 the offending event is defined as "an alcohol concentration of 0.08 or higher as shown by analysis of the person's breath or blood", and RCW 46.61.506 says that breath analysis is based on "grams of alcohol per two hundred ten liters of breath", and Ohio's ORC 4511.19 is more detailed and states the threshold as
concentration of eight-hundredths of one per cent or more but less than seventeen-hundredths of one per cent by weight per unit volume of alcohol in the person's whole blood
concentration of ninety-six-thousandths of one per cent or more but less than two hundred four-thousandths of one per cent by weight per unit volume of alcohol in the person's blood serum or plasma
eight-hundredths of one gram or more but less than seventeen-hundredths of one gram by weight of alcohol per two hundred ten liters of the person's breath
There are also thresholds for urinalysis – really, the Ohio laws are incredibly detailed and specific. Washington, however, also says that just because they give a couple of specific numbers and methods of detecting driving under the influence, that (minimal) details "shall not be construed as limiting the introduction of any other competent evidence bearing upon the question whether the person was under the influence of intoxicating liquor or any drug" (I don't think urinalysis for alcohol is yet admissible in Washington, though it is part of the statute for drug levels).
The above numbers are set because a scientifically supported correlation exists between alcohol in breath and alcohol in blood (you just have to specify the volume of breath -- 210 liters), or urine. Hypothetically, this could be done with sweat. Sweat analysis via Secure Continuous Remote Alcohol Monitoring (SCRAM) is a possible means of determining a violation (of some sort). A relevant case is People v. Dorcent. This case involved terms of probation for a person who had been convicted, and part of the deal for probation was to not drink alcohol for 30 days. The SCRAM bracelet detects whether the subject has consumed alcohol. At issue is whether the method is reliable (and for what purpose). It appears that the method is not up to the scientific level of breath, blood and urine analysis, but it is still acceptable evidence, and is used in most states in the US. One problem is that there is about a 4 hour lag between sweat and blood; it only gives a three-way distinction in levels of non-zero consumption. It is certainly acceptable evidence to establish a violation of a no-alcohol order, and it would be interesting to know if the device has ever been used as the primary evidence for a driving-impaired conviction. It appears that the device has to monitor the subject for a period of hours in order to obtain reliable measurements, so it would be useless compared to other methods of analysis.
Although Washington statutes do not address sweat-analysis as evidence of DUI, since the underlying offense is DUI and not just amount of blood alcohol, it is conceivable that a court could find a "large amount of alcohol" measurement from sweat analysis is sufficiently reliable to support of a conviction under a more stringent BAC law (such as minors or professional drivers) – as long as the court holds that it is competent evidence.
The British courts (and hence police) prefer to measure the alcohol level in a blood sample, as it is less open to challenge for fluctuation or other unreliability: the legal level is 80 milligrams/100 milllitres in England and Wales, and less in Scotland. If a blood sample is impractical for some reason, measurements can be taken from a breath or urine sample; the limits are set out on the Government website. Sweat samples have never been accepted in Britain, nor in any other country that I know of.