“You can’t trust water: Even a straight stick turns crooked in it.” So said W.C. Fields, and that is a lesson for food testing too. Analysis to determine the water content of food can be terribly tricky without a little thought and consideration.

This may seem like a strange assertion to make. Surely moisture testing is one of the easiest tests? People start working in labs by doing moisture tests, don’t they? Don’t you just dry a sample in an oven? Well, all this is probably true. However, let us look just a little bit further.

The first clue to this unlikely complexity is that there are about 20 different methods for the determination of water content in food on the British Standards website. It is most likely that some of the methods are very similar, and a few have been withdrawn. Nonetheless, there are about 20 methods listed. The second clue is that many of these methods are matrix dependent: meat, oilseed residues, spices, coffee, cereals, pulses and so on. What is it all about?

Let us consider a basic method for determination of moisture, and one that has already been alluded to: drying a sample in an oven. Water will evaporate from the test portion and therefore any loss of mass must be equivalent to the moisture content. However, even such a simple process is actually very complicated. Other volatile non-water species such as acetic acid (found in vinegar) may also evaporate, and volatile flavour compounds could be driven off too. Thermally unstable compounds may break down, perhaps even charring in the process. Some water may be very tightly bound in the sample material at a molecular level, and will not be driven off the test material without higher temperature drying. Some compounds may oxidise over the drying time and actually increase their final mass. The test material may even “cook” and trap moisture within its structure.

Therefore, it becomes very easy to appreciate how parameters such as oven temperature, time of drying, the environment within the oven, and presentation of the test material must be clearly defined and controlled. These parameters must be selected in order to deal with the challenges presented by differing sample matrices. This leads to the scenario of drying methods for determination of moisture carrying marked differences in conditions. An example is as follows; classically, meat samples require a fan oven set at 103 degrees Celsius with the test material mixed in with sand and a drying time of a number of hours, whereas a sample of baking powder demands a desiccating vacuum at no more than room temperature.

It may be possible to obtain differing, apparently precise data for differing conditions. If meat samples are dried at 103 °C for 4 hours then they may give slightly different results than if they are dried for 16 hours. Generally speaking, it is considered best practice to keep the drying time to a minimum to prevent chemical changes in the test material. However, there is then a risk of incomplete drying, particularly in a large oven containing many wet samples. Therefore, it may be necessary to carefully balance the applied conditions.

Adding to this complexity is the fact that there are alternatives to physically drying a sample in order to determine the moisture content. Water may be determined by titration, classically using the Karl Fischer titration and applying it to samples of low water content such as confectionary. It may be determined by distillation, which tends to be proposed for herbs and spices. There is even a further step, which is to use instrumentation such as nmr or infrared, although these inference procedures usually require calibration based upon directly analysed data from drying, distillation or chemical techniques.

The next question to be addressed is one too often overlooked. It should be thought through by both food manufacturers submitting samples as well as the laboratories performing the testing. It concerns the purpose of the testing. That is to say, what is the result to be used for? It is at this point that all concerned must consider the fitness for purpose of the method to be applied. If a result is to be used for the presentation of food labelling data then an absolute method uncertainty of ±0.5% (which for most foods is still only a relative value of less than 1%) would be most satisfactory. In point of fact, the vast majority of moisture testing would fall into such a bracket and for which basic, rugged testing methods are perfectly adequate. However, in a production control environment then this might well be an unacceptable level of uncertainty. This is particularly true for products with low levels of moisture such as crisps or biscuits. It may well be that slightly different conditions will need to be applied in order to obtain data with appropriate levels of precision for specific samples.

The final question is one that is relevant to all food testing; that of sample preparation or homogenisation. This can be particularly significant for resistant test material, or materials that take quite a lot of effort to homogenise. Unfortunately, the first law of thermodynamics will come into play and as more work is done to a system then the temperature will usually increase. Therefore, volatiles such as moisture may well suffer from evaporation during an extended or vigorous grinding process. It is not possible to over-emphasise the importance of the correct and controlled homogenisation of test material within food analysis.

In summary, although moisture testing is fairly standardised, there really is no such thing as one standard moisture test. One thing is certain, which is that the test material and the purpose of testing should both be considered before a method is selected, as it is likely to be the applied method that will actually determine the result obtained.