Analysing a sample of food for its fat content can be very simple or it can be very complicated; it can require highly skilled analysts or the most basically trained operative; it can be the bane of the analytical chemist’s life or a triumph of their art. One way or the other it can be considered as a lot of “fun”, regardless of how you define the word.
There are a number of different issues associated with fat testing, but two that inevitably lead to much of the fun are the beautifully paradoxical –
- fat is a bit tricky to define
- there are lots of different ways of finding it
So, how come we are not too sure what we are looking for but we have lots of ways of doing it?
Well, of course we do pretty much know what we are looking for. We even have food legislation that defines fat for us. The EU Food Information to Consumers Regulation defines fat as total lipids, and includes phospholipids. Therefore, fat really means fatty acids, neutral fats, glycerides, compound lipids (glycolipids, phospholipids etc.), waxes and sterols; not one thing, but many, and a mixed bag at that.
I have to admit that I rather like definitions of fat that talk about anything that is soluble in non-polar organic solvents and that is, along with protein and carbohydrate, a main constituent of living cells. I like this definition because I am an analytical chemist, and I now have something to use as a way of selectively defining my analyte of interest. It is only fat that holds this property of solubility in those organic solvents. I don’t really care if it is any of the different compounds listed in the previous paragraph. I do care if it will dissolve in a non-polar solvent, and if it does I am going to claim it as “fat”.
Having defined the analyte, we are now faced with the question of how the fat is exposed to the solvent. The vast majority of food materials contain water, and most of those contain significant proportions of water. In addition, some of the fat may be physically bound into the food material itself. This will all interfere with our extraction process. However, if we in turn interfere with our sample to resolve this issue then might we not affect our results in some way? In addition, might some solvents be better to use than others?
And that I suppose is the key to it all – the fact that different approaches may end up giving different results because they may do different things to the sample. The determination of fat in milk using the Rose-Gottlieb method is almost an art form, and a skilled analyst may achieve remarkable levels of precision. However, if that method were applied to minced bacon then I would be very wary of the results. Conversely, a good old-fashioned acid-hydrolysis approach that is perfect for bacon might leave the milk a quivering mess. This is not to say that happy mediums are not to be found, but that they have to be applied carefully, and with thought, skill and technique.
The determination of fat content is therefore challenging. Methods are usually multi-step processes in which much can go wrong. Poorly executed fat determinations will almost always give low results as the solvent fails to extract the fat, for whatever reason or reasons. Unlike analysis for moisture content, not only is it that the sample itself governs the process applied, but that the analyst is almost certain to be a significant source of method uncertainty as well.
One final thought concerns inference methods, such as IR or nmr. I am a great fan of nmr technology and would suggest that it is an excellent solution for the routine testing of food. Indeed it is a preferable approach for many laboratories. However, for the most accurate and precise measurements of fat content, we may still have to turn to our more traditional methods that can deal with the fact that we are never exactly sure what we are looking for.