Table of Contents

Proximate composition
Use of compositional data
Accuracy of analysis
Preparing and analysing the sample
Choice of method
Stubbs & More procedure


This note describes methods of measuring the main classes of substances present in fish and fish products, namely proteins, water, fat, minerals and carbohydrates. Descriptions of the methods are preceded by more general information on the uses that can be made of compositional data, on sampling of the material being analysed, preparation of a sample, and the accuracy of analysis.

The methods described are those used at Torry Research Station specifically for fish; some are already embodied in national compilations of recommended methods, some are modifications of recommended methods, and others are faster or simpler, but perhaps less accurate, methods often required in the fish industry.

It is not possible in this note to give detailed advice on laboratory practice and safety. Some of the methods are simple and safe enough to be used by relatively inexperienced personnel, but others that involve the use of strong acids or alkalis, high temperatures or organic solvents should not be tackled by inexperienced persons. All of the methods are given in sufficient detail for a trained analyst to use them.

Proximate composition

Composition described in terms of classes of substances present is generally referred to as proximate composition, because the classes or groups, for example proteins or minerals, are those first arrived at in the process of analysis; in proximate analysis the groups are measured as such, rather than as individual proteins or specific minerals. The word proximate in this context does not refer to accuracy; it does not mean either exact or approximate.

The five classes of substance for which methods are given are proteins, water, fats, minerals and carbohydrates. The measured weight of each class is usually stated in the singular, for example as protein content or fat content, and expressed as a percentage of the original weight of the sample. Less commonly, the measured weight of each class is given as a percentage of the dry weight or of the dry fat-free weight of the sample.

Other terms are sometimes used for the classes of substance, for example moisture instead of water, oil or lipid instead of fat, and ash instead of minerals, but to avoid confusion their use is largely avoided in this note.

Use of compositional data

The most common application in industry of the measurement of composition is in quality control; simple and rapid methods are required that can be applied to fish or fish products as they pass through a processing factory. It is seldom necessary to analyse a product completely; often the measurement of only a single component is required. Examples are measurement of water content of fish meal leaving the dryer, of fat content of herring or mackerel to assess suitability for smoking, and of protein content of fish blocks to estimate their true fish content. Nutritionists and dieticians need to know the proximate composition of all foods in order to estimate the intake of the principal nutrients in the human diet and to calculate energy values of different foods. Extensive compilations of the composition of foods are available, derived from proximate analyses made over a long period. These tables also include details of individual substances such as minerals and vitamins. Regular updating of the tables is required because of changes in the nature and source of raw materials and in manufacturing processes.

Measurement of fish content of manufactured products is often necessary to ensure that they meet the requirements of food regulations or commercial specifications. For example, fish cakes must by law contain at least 35 per cent fish, and coated fish products made by most manufacturers contain at least 50 per cent fish, although there are at present no legal requirements in the UK. The analytical procedure for determining the fish content of products that include other constituents begins with a proximate analysis of the mixed product. The whole procedure, called after its originators Stubbs & More, is described in more detail later, because of its importance to manufacturers and consumer protection authorities.

Fishery byproducts such as fish meal require analysis, because animal feeding stuffs containing them have to carry a statement of composition; protein content always has to be given, and usually fat content.


No two fish are identical in composition, and parts of a single fish are likely to be different. Variability poses a particular problem in measuring fat content. It is important, therefore, to be aware that a particular analytical result applies only to the material analysed. Sometimes the composition of a single fish is of interest, but more often the average or representative composition of a large batch of fish or fish product is sought. Most methods of analysis destroy or alter the material so that it can no longer be used for its intended purpose; thus in practice it is necessary to take a sample of the hulk to be analysed. The sample must be large enough to ensure that the analysis is representative, but not so large that material is wasted.

The size of sample will depend on the variability within the bulk material of the component being analysed, and on the level of accuracy required. For example, if the likely degree of variability of fat content of a batch of mackerel is known from previous experience, one can estimate statistically the size of sample that when analysed will give an average fat content for the batch with an error of say less than 5 per cent of the determined value. But prediction of sample size based on experience is not always possible; some sampling plans have been published, for example for prepacked frozen products and for fish meal consignments, which can be used as sampling guides, and general advice on sampling procedure can be found in textbooks on quality control. It must be emphasized that analysis of one fish or even a few fish will seldom give a reliable indication of the composition of a large batch.

Accuracy of analysis

Each of the components in a sample is present in a precise quantity, but analysis may not give the true value, for a number of reasons. First. no measuring procedure is free from error; an analyst measuring two successive identical samples by the same method can produce slightly different answers, because of random errors arising from minor differences in procedure at each stage of the analysis. Wherever possible, a method with a low intrinsic error should be chosen.

Secondly, the method may not measure exactly the quantity sought. For example some water in fish is loosely held, and can be squeezed out by light pressure, while some is strongly bound to protein; measurement of water content will thus depend on the degree to which the method can remove bound water. Drying in dry air at room temperature will give a much lower apparent water content than drying at a temperature above 100°C, and even at the higher temperature the answer will depend on the duration of the heating period. To measure total water content would require a lengthy period at over 100°C, and such a method might prove too time-consuming. Again, other weight changes may occur during heating, such as the loss of other volatile substances and the uptake of oxygen from the air, so that the total weight change does not entirely represent water loss. It is therefore important to describe the method of analysis when quoting the results so that, when comparisons of results are made, any differences in methods can be taken into account.

Thirdly, variability within a sample can affect the results; no matter how well a sample is mixed it is unlikely to be absolutely homogeneous, and two successive portions taken from it for analysis will inevitably differ to some degree.

Preparing and analysing the sample

Once a sample has been selected, it may consist of a number of items, and it has to be decided whether to analyse each item in the sample separately, or to combine them and make a single analysis. Separate analysis of each item gives information about variability from the mean result within the sample and hence within the batch from which the sample was drawn. For example, where a sample consists of single items drawn regularly from a continuous processing line, separate analysis of each, using a rapid method, will indicate any consistent or increasing deviation from specification, and allow prompt remedial action to be taken. On the other hand, if the average fat content of a large batch of herring is required, it is cheaper and quicker to combine all the fish in the sample and make a single analysis.

Each sample or part of a sample that is to be separately analysed must be made as uniform as possible before measurement begins. The material, whether a single fillet, several whole fish or a number of packs of product, must be thoroughly mixed to reduce variability before a portion is taken for analysis. Fish and fish products should be passed twice through a mincer with holes not more than 4 mm in diameter, and the mince then thoroughly mixed. A domestic food processor or a commercial-scale grinder and mixer can be used to comminute and mix the material at the same time. The mixed sample should be kept at or below 4°C in a closed container. Some separation of components, especially fat, can occur on standing, so the sample should again be thoroughly mixed before being used.

Fish meal should be ground to pass through a 1 mm mesh, and then thoroughly mixed. Portions should be taken quickly for analysis in order to keep any change in water content to a minimum.

The analysis for each required component of a mixed sample should be done at least twice, and the results should not differ by more than 2 per cent of the mean value. If they do differ, two additional analyses should be made and, if there is still no clear agreement, the analyst should repeat the procedure on material of known composition. If the known sample gives a satisfactory result, it is possible that the unknown sample is still not uniformly mixed, but if the known sample gives an unsatisfactory result, the analyst may need further advice. The mean of the two or more concordant results should be given in the final report of the analysis, together with a brief description of the method used.

Choice of method

Recommended methods for the analysis of fish and fish products were published by the Analytical Methods Committee (AMC) of the Royal Society of Chemistry in 1979. Other compilations of methods exist, notably the Recommended Methods of the Association of Official Analytical Chemists (AOAC) published in the USA. Some standards, regulations and codes of practice specify particular analytical methods, and these should be used when checking compliance with such requirements.

The methods described below include methods recommended by the AMC, with occasional modifications, and other methods that are simpler or more rapid.


Protein is usually determined by measuring nitrogen, the characteristic element in protein, rather than protein itself; estimation of protein directly is a more time-consuming procedure. Not all substances containing nitrogen, however, are proteins, so the quantity estimated from measurement of nitrogen is usually called crude protein, which in addition to true protein includes free amino acids, trimethylamine oxide and its decomposition products, and other substances. The nitrogen content of many proteins is about 16 per cent, and so the nitrogen content of a sample of fish is conventionally converted to crude protein by multiplying by 6·25. Some species of fish such as skate and dogfish have a high urea content and, since urea has a high nitrogen content, the crude protein content of such fish can give a misleadingly high estimate of true protein.

Kjeldahl method

The following procedure is a slight modification of the AMC method; full details of the principle of the Kjeldahl method and the apparatus required can be found in most textbooks of analytical chemistry.

Place 15 g of potassium sulphate and 0·5 g of copper (II) sulphate in an 800 ml Kjeldahl flask. Weigh accurately about 2 g of the prepared sample (1·5 g if rich in fat, 0·5 g if fish meal) on a filter paper and transfer both to the flask. Add carefully 25 ml of concentrated sulphuric acid and mix by swirling the flask.

Place the flask in an inclined position on a suitable heating device in a fume cupboard. Heat carefully until foaming has ceased and the contents have become liquefied. Digest by boiling gently, occasionally rotating the flask, until the liquid is completely clear and of a light blue colour; boil gently for a further 1½ hours. Total digestion time should be not less than 2 hours. Cool the contents of the flask to about 40°C and cautiously add 50 ml of distilled water. Mix and allow to cool.

Transfer the contents of the flask to a 250 ml standard flask, rinsing several times, and make up to 250 ml. Transfer 50 ml to a suitable steam distillation apparatus. Put 25 ml of 4 per cent boric acid solution in a conical flask and place under the condenser so that the outlet dips into the liquid. Add 35 ml of 33 per cent sodium hydroxide solution to the distillation flask. Steam distil for 4 minutes after the first drop of distillate, or until the distillate is no longer alkaline. Lower the conical flask so that the condenser outlet is above the liquid level and distil for a further minute. Add to the flask 4 drops of indicator (0·2 g methyl red and 0·1 g methylene blue in 100 ml ethanol) and titrate to a grey colour with exactly 0·1 M hydrochloric acid. Repeat the analysis with a second 50 ml portion of digest.

A complete blank determination using only a piece of filter paper instead of the fish should be carried through regularly.

Nitrogen content is calculated by:


V1 is the mean volume in ml of 0·1 M hydrochloric acid required for fish,

V0 is the mean volume in ml of 0·1 M hydrochloric acid required for blank, and

M is the weight in grams of the portion taken of the sample.

Nitrogen is finally converted to crude protein by multiplying by 6·25.

There are many variants of the Kjeldahl procedure. Automatic or semi-automatic systems for this analysis are available; they are expensive, but are rapid, safe, and generally accurate, using either an established procedure like that above, or a procedure recommended by the makers.

Other methods

Many direct and indirect methods of measuring protein have been devised over the years. A method of current interest uses measurement of the infrared absorption by a finely divided suspension of the food in water; by measuring at a number of selected wavelengths it is possible to estimate separately the protein, water, fat and carbohydrate contents. A number of automatic instruments are now available for this technique.


Water content is usually measured either by drying in an oven and measuring the weight loss, or by the Dean & Stark volumetric method.

Oven drying

Weigh accurately, after drying at 100°C and cooling, a suitable container, of aluminium, glass, silica, platinum or stainless steel, with a loose fitting lid. Spread out 5 g of the prepared sample evenly over the bottom of the container, put on the lid and weigh accurately again. Place the container in an air oven thermostatically controlled at 101°C, remove the lid and leave for 24 hours. Replace the lid, transfer the container to a desiccator and allow to cool to ambient temperature. Weigh the closed container.

When drying samples with a high fat content, it may be necessary to prevent accidental weight loss by covering the fish with a filter paper to stop droplets erupting out of the container. The weight of the filter paper must (not much) be included with that of the container. When measuring fish meal by this method, weigh the filled container as soon as possible after loading. Otherwise the procedure is the same.

A vacuum oven can be used in place of an air oven when it is necessary to minimize the effects of thermal damage. If drying is done for 24 hours at 70°C and the rest of the procedure remains the same. the results are directly comparable with those obtained at 101°C in an air oven. The main disadvantage of the vacuum oven, apart from greater cost, is that it is difficult to achieve uniform temperature distribution throughout the oven.

Infrared ovens with an integral weighing facility are available that can be used for rapid determination of water content; drying times of 1-2 hours are typical for fish products. There is a significant risk of thermal damage to the material, and only one sample can be measured at a time. Some microwave ovens can also be used to give drying times of 1-2 hours, and a number of portions can be handled simultaneously, but there remains the risk of overheating. Microwave drying is not suitable for relatively dry products like fish meal.

Dean & Stark method

Weigh accurately about 10 g of the prepared sample (25 g of fish meal) in the 250 ml round bottomed flask of a Dean & Stark apparatus and add about 100 ml of toluene. Fit the receiver and condenser to the flask and add toluene to the condenser until the receiver is full. Heat the flask until smooth but vigorous boiling is achieved. Continue boiling for about an hour or until water has stopped accumulating in the receiver. Any droplets remaining in the condenser or the upper parts of the receiver are dislodged into the receiver by using a rubber tipped glass rod and a stream of toluene. Allow the receiver to cool to room temperature and measure the volume of water.


Most methods of measuring fat content depend on extracting the fat by dissolving it in a suitable solvent. In the method given below the fat is recovered from the solution by evaporating the solvent and is then weighed.

Modified Bligh & Dyer method

Weigh accurately 50 g of the prepared sample on a watch glass. Transfer the fish to a blending jar, using a little of the solvent required in the next stage to wash any remaining fragments from watch glass to jar. Add 160 ml methyl alcohol, 80 ml chloroform and enough water to make the total present in the jar up to 64 ml. If the water content of the sample is not known, this must first be measured fairly accurately in order to determine the amount of water to be added; for example, if the water content of the sample is 60 per cent, then the 50 g of sample contains 30 ml of water, and 34 ml water must be added to make up the required 64 ml. The ratio of water: methyl alcohol: chloroform must always initially be 4:10:5.

Blend the mixture for 2 minutes. Add a further 80 ml chloroform and blend for 1 minute. Add 80 ml water and blend for 30 seconds.

Filter the contents of the blending jar through a Buchner funnel into a Buchner flask. Wash down the blending jar with a little chloroform, and transfer the wash to the funnel. Decant the filtrate from the Buchner flask to a 1 litre separating funnel, washing out the flask with a little chloroform. Allow the filtrate to settle into an upper aqueous layer and a lower chloroform layer. When the layers have clarified, run off the chloroform layer into a 1 litre round bottomed flask. If the chloroform layer takes too long to clarify, run it off through a filter funnel containing a small amount of anhydrous sodium sulphate to remove traces of water. Wash the filter paper and sodium sulphate with chloroform. Heat the flask at 35-40)°C on a rotary vacuum evaporator to evaporate the chloroform. When nearly all the chloroform has been evaporated, transfer the contents of the flask to a 100 ml round bottomed flask that has been dried and weighed. Heat the flask on the rotary evaporator. Remove the last traces of chloroform under high vacuum or in an oven at 100°C until the weight of the flask is constant.

Fat content (%) = final weight of flask contents in grams × 2
By this method, more than 95 per cent of the fat content of the sample should be successfully extracted and measured.

Other methods

Proprietary apparatus is available in which fat is extracted from a known weight of the prepared sample using tetrachlorethylene, and the solution of fat, after filtration, is transferred to a float chamber where its specific gravity is measured. Conversion tables are used to translate specific gravity to fat content. Although the apparatus is expensive, it is easy to use, and the method is fast. Equally simple infrared absorption methods may be used.


Minerals, the inorganic components of fish, often collectively called ash because of the method of measuring them, are seldom of direct technological interest. But their measurement is an essential part of any total proximate analysis, and is part of the Stubbs & More procedure for composite products described later. Measurement of ash is sometimes a useful indicator of the amount of leaching of soluble constituents of fish resulting from contact with water or melting ice.


This is the AMC method in which the material is heated to a high temperature to drive off all water and volatile substances, and to decompose all organic matter.

Heat a silica or platinum dish in a furnace at 550-600°C for 20 minutes. Remove and cool in a desiccator. Weigh the dish accurately, add about 5 g of the prepared sample, spread it out evenly, then reweigh accurately. By pipette, add exactly 1 ml of magnesium acetate solution (25 g of anhydrous magnesium acetate made up to 100 ml in water) to the dish, distributing it as evenly as possible over the material. Dry and char the portion by heating carefully, then heat at 550-600°C in the furnace for 3 hours. The initial drying and charring may be done by heating with a gas burner, or by putting the dish in the cold furnace and heating carefully to the final temperature. Cool in a desiccator and reweigh. Reheat for 30 minutes, cool and reweigh, and continue until successive weighings agree. In practice, heating overnight is a convenient procedure. Carry through a blank determination regularly, using magnesium acetate solution only, and subtract from the measured weight of ash.

It is often possible to measure water and ash in the same portion; if a suitable dish is used for the determination of water by drying, magnesium acetate solution can then be added and the contents of the dish ignited.

Acid-insoluble ash is sometimes measured as an indication of contamination, especially in fish meal; the equivalent term sand is sometimes used. To determine this quantity, the total ash is transferred to an evaporating dish with dilute (3 M) hydrochloric acid. Evaporate to dryness and heat for a further hour to dehydrate silica. Resuspend the residue in 50 ml of dilute hydrochloric acid, bring to the boil and filter while warm through an ashless paper. Wash the residue repeatedly with warm water until the filtrate is no longer acidic. Transfer the filter paper to the original dish and repeat the total ash procedure.


Fish muscle normally contains only traces of carbohydrates, in the form of sugars, sugar phosphates and glycogen. Some other tissues such as liver contain larger amounts as glycogen, and most molluscan shellfish also contain a fair amount of glycogen. There is no single method suitable for determining total carbohydrate in all tissues and, apart from the indirect infrared method mentioned earlier under protein, the methods are not straightforward. For these reasons it is common to estimate carbohydrate (C) by difference.

C (%)=100-P-W-F-A
where P is percentage protein (nitrogen × 6-25)
W is percentage water
F is percentage fat
A is percentage ash
Because of possible accumulation of errors in the four separate analyses, the accuracy of such an estimate may be low, especially if the amount is small.

Stubbs & More procedure

Some manufactured fish products contain large amounts of carbohydrate, for example as cereal filler in pastes and spreads, as potato starch in fish cakes, or as wheat flour in battered and breaded products. The carbohydrate content of a prepared mixed sample can be estimated by difference as indicated in the previous section. The Stubbs & More procedure is an accepted means of using all the analytical information to derive an estimate of the actual fish content of the product. The total crude protein, obtained by multiplying the total percentage nitrogen by 6·25, is the sum of the crude fish protein and the protein contributed by the flour or other high carbohydrate ingredients. To exclude the latter contribution the carbohydrate content is multiplied by an agreed factor, 0·02 for wheat flour, to give an estimate of the nonfish nitrogen. Subtracting this value from the total nitrogen content gives the fish nitrogen, NF. The fish content is then

The value to be used for the nitrogen factor in this expression is the mean value of the percentage nitrogen content of the species of fish used. Recommended values have been derived by the AMC for cod (2·85) and saithe (2·90) on the basis of extensive surveys of the nitrogen content of these species throughout the year and over a range of fishing grounds. Values for other species are available, but the reliability of the figures is less certain than for the species mentioned.

This procedure is clearly too time-consuming and complicated for routine quality control purposes, and simpler methods of measuring fish content of frozen coated products have been devised. These methods simply strip the coating, after partial thawing, from the still frozen fish core, which is then weighed. Many such methods have been proposed which differ only in detail and which give results that are generally in agreement with those obtained by the Stubbs & More procedure.

If you have any enquiries, write, 'phone, or call at the address below:

The Director,
Torry Research Station,
PO Box 31,
135 Abbey Road,
Aberdeen AB9 8DG
Tel. 0224 877071
Other recent Notes in this series, which are available free of charge in the UK from the above address are:
61 Gaping of fillets, by R. M. LOVE.

62 The freezing time of fish, by F. J. NICHOLSON.

63 Fishing ports in the UK, by J. J. WATERMAN.

64 Fish silage, by I. TATTERSON and M. L. WINDSOR.

65 Fishworking machinery, by S. MAIR.

66 Handling and processing mackerel, by J. N. KEAY.

67 The haddock, by J. J. WATERMAN.

68 Icemaking plant, by J. GRAHAM.

69 Cook-freeze fish products, by J. N. KEAY.

70 Advice for the fish industry; who does what, by J. J. WATERMAN.

71 Processing cod; the influence of season and fishing ground, by R. M. LOVE.

72 Reducing odour in fish meal production.

73 Stowage of fish in chilled sea water, by J. H. KELMAN.

74 Handling and processing rainbow trout, by A. MILLS.

75 Freezing small pelagic fish, by I. McDONALD.

76 Dark colour in white fish flesh, by R. M. LOVE.

77 Squid, by G. D. STROUD.

78 Health hazards of handling industrial fish, by A. WARD.

79 Minced fish, by J. N. KEAY.

80 Round worms in fish, by R. WOOTTEN and D. C. CANN.

81 Handling and processing blue whiting.

82 Hot smoking of fish, by A. McK. BANNERMAN.

83 Fish smoking: a dictionary, by J. J. WATERMAN.

84 Handling and processing oysters, by G. D. STROUD.

85 Chilled and frozen fish: a dictionary, by J. J. WATERMAN.

86 Shopping for fish: advice on quality, by A. CRAIG.

87 Composition and quality of fish: a dictionary, by J. J. WATERMAN.

88 Packing fish in a modified atmosphere, by D. C. CANN.

Earlier notes in the series most of which are still available, are summarized in:

60 Key to Advisory Notes 1-59, by J. J. WATERMAN.

Crown Copyright Reserved

Printed by HMSO, Edinburgh Press
Dd 8823145 10M 4/84 (215228)

Top of Page