Assuming we are talking beer here: Refractometry does work and can be used in operations where the same beer is brewed time and time again. But the refractometer must be calibrated using one of the more absolute methods. Thus refractometry is great for telling you that this batch is 0.2% higher than that batch of Old Overshoe but not so great for telling you what the ABV of the first batch of Old Overshoe is. You would send a bottle of OO off to the lab to get the ABV measured and when you got the number build the ABV vs RI curve. ASBC has a Method of Analysis (MOA) based on this.
To measure beer accurately one must separate the alcohol from the beer by distillation then measure the density of the distillate. The former step requires some familiar lab gear (mantle, Kjeldahl adapter, condenser, receiver, volumetric flask) for the distillation part. Total investment of a couple of hundred $. It's in measuring the density of the distillate that you jeopardize your children's college educations. Tralle hydrometers don't work because the density of an alcohol/water solution at the strength of typical beer is so close to 1 that he marks on the instrument are too close together to read accurately. The way to go is a digital density meter. Put the sample in, push the button, read the ABV (or ABW, OIML or AOAC...) but these are out of the question even for fairly large breweries. An appreciably less expensive option is gas chromatography.
An affordable option is the simple pycnometer. Equipped with one of those and good analytical balance you can measure the density of the distillate and calculate the ABV. But this requires a good balance and while that is less expensive than a GC or digital density meter it is still a pretty expensive proposition.
For dry wines and meads ebuliometry and surface tension (Vinometer) are workable and for wines, meads and higher alcohol beers a Tralle hydrometer can be used.