A.D. Evers, Ascus Ltd. Talybont, Albert St, Markyate, Herts, England.

M Kelfkens, TNO Nutrition and Food Research,PO Box 360,Utrechtseveg 48,3700 AJ Zeist, The Netherlands.

G. McMaster, BRI Australia Ltd, PO Box 7 North Ryde, NSW 2113, Australia.

You will notice that, although I have the privilege of delivering this paper, two distinguished co-authors are also credited. The content of this talk reflects something of the contributions made by the institutes that we have served for many years, to the understanding of flour quality and measurement of its purity. We have included findings from CCFRA, in the UK (for whom I worked until 1997) TNO in the Netherlands, where Marcel Kelfkens is a Scientific Co-worker and BRI Australia Ltd where Graham McMaster is the Chief Executive Officer.

As an introduction to these contributions we provide a reminder of the importance of white flour in the market, objectives of milling white flour, and some of the methods of assessing the degree of success achieved.

One universal objective of flour milling is the reduction of particle size, or more precisely to grind grains to produce a flour or a meal. Where this is the only objective the resulting product is a wholemeal in which, by definition, the botanical constituents are present in the same proportions as in the wheat from which it is produced. While wholemeals can be baked to produce excellent products they are, in many countries, less popular than products made from flours that are more refined. The miller thus has a further objective: the separation of nutrients originating in the starchy endosperm from those originating in other parts of the grain, i.e. the embryo and the seedcoats and fruitcoats. In the United Kingdom wholemeal and brown flours account for only about 14% of the bread flour market (Anon 1999). The majority of remaining flours are described as ‘white.’

While defining wholemeal is simple, defining brown and white flours is much more difficult as a wide range of products can be made within both categories. However, the distinction between the two classes of flour as ‘brown’ and ‘white’ demonstrate that colour is an important recognisable feature by which they can be distinguished, and this is possible because starchy endosperm is virtually colourless while other components are more pigmented.

It is not surprising that products made from darker flours are themselves darker in colour and in some cases this explains their lesser popularity. However the main reason for whiter products being preferred lies in the fact that the functional components that give wheaten flour products their gas retaining properties and hence lighter texture lie within the starchy endosperm. Hence in whiter flours the functional components are more concentrated. When embryo, seed coat and fruit coat fragments are present they dilute the functional components and reduce gas retention. In producing white flours therefore, the miller seeks to produce ground endosperm with as few particles as possible originating from other grain tissues. Because these tissues are more pigmented than endosperm the miller has traditionally relied, during production, on visual examination to judge how successful he has been. The value of visual assessment is limited because it is subjective and for purposes of trading, alternative methods such as ash determination and colour reflectance measurement have been adopted. In recent years advances in computer technology have made it possible to provide, by image analysis, the advantages of visual examination without its shortcomings. In other words quantification by image analysis is relevant, objective and repeatable. In this paper, because of time limitation, we are concentrating only on ash and image analysis methods.

Let us examine the morphology of the grain. Fig 1: Diagram of LS of wheat grain.

wpe4.gif (54509 bytes)

and the contribution that individual components make to the whole.

wpe3.gif (10578 bytes)

Fig 2 pie chart of contributions to dry mass.

The starchy endosperm is a relatively simple and uniform constituent contributing over 80% of the grain dry mass. The embryo is much smaller, it comprises the embryonic axis, and the scutellum.

The remaining components all partially or completely surround the endosperm and embryo. They consist of fruit coats or pericarp, seed coats (testa), nucellus and aleurone. The individual layers and their significance in the context of measurement by ash value and image analysis will be discussed later.

If the milling of white flour could achieve perfection, it would quantitatively separate starchy endosperm from all other components. The fates of all components in such a perfect system are shown in Fig 3

wpe5.gif (34465 bytes)

Fig 3 Relationship between grain parts and milling fractions.

Fragments of non-endosperm components are undesirable in white flour for at least three reasons:

  1. they are non-functional and hence dilute the functional components present in starchy endosperm,
  2. dark specks are visually unattractive and in some products may be perceived as fungal colonies,
  3. in aerated products they interrupt the continuity of gas cell walls, causing small holes and thus reducing gas holding capability.

Let us now consider how ash measurement and image analysis provide an indication of flour purity.

Ash determination consists essentially of burning a known mass of the material to be analysed and measuring the residue. As burning destroys all but the mineral components the mass of the residue provides an indication of the contribution that minerals made to the original material. The application of the method to determining bran content of flour is justified by the fact that endosperm has a lower mineral content than bran. Fig 4 shows the contributions of the grain components to total ash. Because of the large contribution that endosperm makes to the grain mass it contributes 14% of total ash.

wpe6.gif (7317 bytes)

Fig 4 Pie chart showing contributions of grain parts to total ash.

Image analysis depends entirely on colour differences between components and enjoys the advantage that the contribution for starchy endosperm to coloured components is zero.

A thoughtful milling engineer (E.D.Simon) wrote in 1928:

The whole problem of the miller comes down to this, to break up the endosperm without producing at any stage any bran powder. Once bran powder is produced it goes into the flour and can never be got out again. The unforgivable sin for any milling machine is the production of bran powder. If this is a correct statement of the case, then clearly the first condition of progress is a proper test for the presence of bran powder. No such test exists today.’ The whole problem of the miller comes down to this, to break up the endosperm without producing at any stage any bran powder. Once bran powder is produced it goes into the flour and can never be got out again. The unforgivable sin for any milling machine is the production of bran powder. If this is a correct statement of the case, then clearly the first condition of progress is a proper test for the presence of bran powder. No such test exists today.’

It is worth noting that the ash test was in use in 1928 but Mr Simon did not consider it suitable. The wisdom of his assessment has been endorsed by recent studies at TNO where scientists were investigating the relationships between bran content and flour performance in baking tests and other quality tests. The study was carried out over a period of 3 years involving 21 varieties of wheat. A total of 120 samples were milled, tested and baked. The correlation between ash values and performance parameters did not achieve statistical significance but pericarp contents measured by image analysis did. Clearly ash values were not usefully reflecting either the amount of bran present nor any other value capable of being used to predict performance quality.

Fig 5 Correlation coefficients between quality parameters and predictive test results. [3 years 21 varieties, 120 samples]


LOAF VOLUME n.s. -0.71
ZELENY n.s. -0.65
R -max n.s. -0.62
Area extensograph n.s. -0.54

The obvious question arising from these results is why two methods that set out to measure the same thing can produce such conspicuously different results.

TNO scientists compared ash values and pericarp contents of individual flours and it became clear that some flours with low ash values had high pericarp contents and some flours with high ash values had relatively little pericarp present. They looked at the ash values of the parent wheats and found that quite a variation existed. Not only that, it was possible to show a close relationship between flour ash and grain ash. They considered the possible underlying reasons for this and two possibilities occurred to them:

  1. The grains with high ash contained a greater proportion of non-endosperm material
  2. One or more of the botanical components of the high ash grain had higher ash values

wpe8.gif (11400 bytes)

Fig 6 Flour ash values as a function of parent wheat ash values.

It turned out that the second possibility was correct and one of the botanical components contained more mineral residues. While this had been predicted, two aspects of it were surprising:

  1. the botanical component that varied was the endosperm.
  2. The variation was as great as 50% (between 035% and 0.53%). So large was it in fact that it accounted for nearly 80% of the variation in flour ash.

The importance of this shortcoming can be illustrated by reference to data from actual millings of four varieties with endosperm of different ash values. The data are presented as ‘ash curves’ which require some explanation. Flour milling is a multistage process and many streams of flour are produced throughout the sequence of roller milling passages. Individual streams make different quantitative contributions to the total flour and the purity of the streams also varies. Ash curves are constructed by taking account of both these factors in plotting cumulative yield against the corresponding ash value. Thus the first point plotted represents the yield and ash value of the purest of the flour streams. The second point represents the combined yield and ash value of a mixture of the purest and next purest stream. Further points represent values of the mixture as progressively less pure flours are added until all flour streams are included.

Fig 7 shows a family of ash curves comparing results for four wheat varieties. All wheats were milled under comparable conditions on a pilot mill.

spotlight_figure1.gif (11890 bytes)

Fig 7 ash curves for four wheat varieties milled under comparable conditions.

It should be noted that the most significant differences among curves lie at the origins of the curves, indicating that the greatest variation was among ash values of the purest flours from respective wheats. Thereafter, the curves are almost parallel indicating that inclusion of subsequent streams had similar effects in all cases.

The purest flours represent almost pure endosperm and thus differences reflect intervarietal variation in endosperm ash and not degree of inclusion of bran. The curves, being almost parallel, give little indication of differences in milling properties of the four wheats.

Compare this family of curves with a similar set in which bran content replaces ash values

spotlight_figure5.gif (12460 bytes)

Fig 8. bran content curves for four wheat varieties milled under comparable conditions.

Origins of the curves lie much closer together than in the case of the ash curves and the shape of curves is much more variable, indicating different milling properties among the varieties.

It is difficult to overestimate the significance of the challenge that these results present to the validity of ash as a measure of purity. Ash values are determined in order to indicate the level to which non-endosperm components are present and yet 80% of ash variation among the flours in the TNO analysis could be accounted for by variation in the endosperm itself. The industry has to ask itself whether such a standard method can continue be taken seriously.

The ability of ash value to indicate total bran content has thus been seriously questioned. Examination of the performance of ash when it is applied to the non-endosperm components also reveals serious shortcomings, as consideration of the details of mineral distribution shows. The first observation to make is that although endosperm contains lesser concentrations of minerals than all the non-endosperm components

wpe9.gif (9696 bytes)

Fig 9 Ash values of major grain components

the minerals are not evenly distributed within them, as can be illustrated by reference to the bran tissues.

wpeB.gif (35805 bytes)

Fig10 Ash values of individual bran tissues

Thus, aleurone tissue has the highest proportion as indicated by its ash value of 16%. The outer pericarp, or beeswing, has an ash value of only 1.3% - less than one tenth of the value for aleurone!

wpeC.gif (9299 bytes)

Fig 11 proportions of individual bran tissues that would be present in a flour of 0.5% ash (endosperm ash 0.3%)

shows the percentages of individual fractions of bran that might be present in a flour of 0.5% ash milled from a wheat the endosperm of which has an ash value of 0.3 The most noticeable feature is that more outer pericarp – or beeswing bran - can be added than any other fraction. This is significant in quality terms for two reasons:

  1. The beeswing layer is only loosely attached to the underlying layers and it separates easily from them.
  2. the beeswing includes the brush hairs that occur at the tip of the grain. It is the brush hairs, with their needle like structure, that penetrate the cell walls in aerated wheat products and cause them to be leaky, thereby limiting their ability to hold gas (Gan et al 1989)

wpeD.gif (50388 bytes)

Fig 12 brush hairs penetrating air cell

Compare this with the aleurone and other bran layers, which more readily conform to the cell wall shape.

wpeE.gif (74952 bytes)

Fig 13 bran layers in air cell wall

Differences among other layers are of less importance because the adherence among them is much greater so they tend to remain attached unless very finely ground. Furthermore if all these layers are regarded as a composite the amount that could be added approximates to the value calculated for the seedcoat.

We should consider how well image analysis performs in this situation. Image analysis depends upon contrast between light and dark materials. As endosperm is white the most easily detected particles are dark. All bran layers are darker than endosperm so they are all detected by image analysis. Experiments have been performed at CCFRA with the Branscan laboratory instrument applied to biscuit flours containing up to 10% beeswing bran and it is clear from Fig 14 that detection is both sensitive and linear.

wpe10.gif (22863 bytes)

Fig 14 Relationship between bran content measured by Branscan and proportion of beeswing added to a biscuit flour (Unpublished – reproduced by kind permission of Incorporated National Association of British and Irish Millers)

So far we have seen evidence of flours of the same ash value containing different amounts of non-endosperm tissue because of variation in starchy endosperm ash. We have also seen how the relative proportions of different bran tissues present might vary without being detected by ash measurement. While such variations must occur when milling conditions are varied in the conventional process they have best been illustrated in an exercise in which aliquots the same wheat were milled by different processes.

The exercise was carried out in the pilot mill at BRI Australia Ltd. (Moss et al 1998) The comparison was made between flours produced using the conventional roller milling flow and a modified flow in which the first stage was processing by a Satake debranner as used in the Peritec system. Bran content of all the machine flours was assessed by ash determination and two image analysis systems: the Dipix fluorescence method and Branscan. The Branscan system is sensitive to all coloured tissues while the Dipix instrument was configured to maximise sensitivity to aleurone tissue. The results were presented as cumulative curves as described earlier.

The cumulative ash curves for the two milling methods were very similar

bri_1.gif (5667 bytes)

Fig 15 Cumulative ash curves for flours produced by a convention mill flow and by a mill flow incorporating an initial pearling stage

indicating that flours produced by both methods did not differ qualitatively. Curves (Dipix) representing cumulative aleurone content of flours produced by the two milling methods were however quite distinct

bri_5.gif (5649 bytes)

Fig 16 Cumulative aleurone curves for flours produced by a convention mill flow and by a mill flow incorporating an initial pearling stage, indicating greater inclusion of aleurone tissue in the flours milled with pearling included in the flow. By contrast, Branscan curves

bri_3.gif (5796 bytes)

Fig 17 Cumulative bran curves for flours produced by a convention mill flow and by a mill flow incorporating an initial pearling stage], showed that the conventionally milled flours had the greater amounts of coloured bran particles present.

These results are easily explained since the pearling process preferentially removes tissues from the outside of the grain and it is these tissues that are most darkly coloured. Aleurone is less completely removed because it is the innermost of the bran layers. In a conventional milling, bran layers would be less inclined to separate and in consequence the bran particles included would be most sensitively detected by the Branscan, because of colour difference.

In summary, measurement of ash as an indication of starchy endosperm purity in white flours is extremely flawed because:

  1. The mineral content of starchy endosperm can vary by as much as 50% and thus can mask variation in amount of bran specks present.
  2. Among bran tissues, beeswing has the lowest mineral content and although it is potentially the most damaging to cell structure it is the least sensitively detected fraction by ash measurement.. Aleurone on the other hand, which is most sensitively detected, is least likely to adversely affect flour performance.

Image analysis offers a preferable alternative to ash as it discriminates sensitively between endosperm and all coloured impurities, showing greatest sensitivity to those particles which disfigure and potentially affect flour performance adversely.

On the basis of comparisons made in this paper and the rapidly accumulating evidence of reliability and reproducibility of image analysis data, I submit that the time is now ripe for responsible standards organisations to begin the process of establishing image analysis standards for endosperm purity in flours.


Anon. 1999 Bread for success. World Grain March 1999 34 (quoted figures supplied by NABIM London)

Gan, Z, Ellis, P.R., Vaughan, J.G. and Galliard, T. Some effects of non-endosperm components of wheat and of added gluten on wholemeal bread microstructure. J. Cereal Science 10, 81-91.

Moss, R., McCorquodale, J. and Osborne, B.G. Comparison of ash, Dipix and Branscn measurements using flours produced on a pilot mill incorporting a Peritec debranning machine. 48th Australian Cereal Chemistry Conference, Cairs, August 1998

Simon E.D. 1928 Flour milling research





site search by freefind advanced