Cole. C.G.B.* and Roberts. J.J. Department of Food Science. University of Pretoria. Pretoria. 0002. RSA.

Originally published in Meat Science, 45(1). 1997.


Gelatine colour is of commercial and scientific significance and yet there is no nationally accepted method for its measurement. On analysis it was found that light scatter due to imperfect filtration and molecular size (Veis, 1964) were the sources of interference. Colour measurements using a particular set up of the BYK-Gardner Color-View Spectrophotometer were found to measure the colour of molten gelatine solutions from the Bloom strength determination, by reflectance, in agreement with visual colour values (r = 0.97), ascribed in accord with Beer's law. Type A and Type B gelatines (67), with turbidities of < 80 NTU, from a wide range of raw materials and manufacturers were assessed.

A400, Absorbance at 400 nm; Corr, Corrected; DGI, Davis Gelatine Industries; DCU, Davis Colour Units; NTU, National Turbidity Units; L, a, b, Tristimulous colour values.


Commercial gelatine is not colourless in solution but has a colour varying from a very pale yellow to dark amber. There can be no doubt that the colour attribute of gelatine has practical significance in that some 60% of world production is consumed by the confectionery industry (Siebert, 1992). In this industry, the products are very often coloured and it stands to reason that, the less the colour variation in the ingredients, the easier it would be to produce a uniform product. Furthermore, in the minds of most people the lack of colour is associated with purity, hence, pale colour is normally more desirable than darker colour. The importance of gelatine colour is recognised by manufacturers (Hoffmann, 1985; Ottenbacher, 1981; Schreiber, 1977).

In the author's experience there are many "in house" methods for measuring the colour of gelatine in solution by visual comparison of unknown gelatines to various standards both in the gelled state and in the liquid state. The numerical colour values attributed could be arbitrary or in conformity with Beer's law or a mixture of Beer's law and arbitrary values. A problem arises however, when for whatever reason two parties want to agree on a measure of colour. It is obvious that in this case a reference instrumental method for gelatine colour measurement would be invaluable.

The problem with the spectrophotometric measurement of gelatine colour (Cole, 1995) is demonstrated below. The cause has been shown to be interference from three sources:

  1.  Variable light scatter due to the molecular size of gelatin. This can be overcome by reducing all gelatins to a uniform molecular size by enzymic proteolosys.
  2. Variable light scatter due to imperfect filtration of gelatin. This can be overcome by passing the enzyme treated solution above, through a submicron membrane filter.
  3. Colour is the result of light absorbed over the visable spectrum. Some gelatines absorb significant amounts of light right on the edge of the infra-red portion of the spectrum and this upsets the integration of the visable absorbance spectrum. This can be overcome by moving the origin of the absorbance spectrum to the value of the absorbance at 700 nm.
Based on the above, a method was developed which allowed a satisfactory measurement of gelatine colour using a modern spectrophotometer, but the manipulations were complex and the method was not considered suitable for routine application.

Saunders and Ward (1953) made suggestions for an instrumental measurement of gelatine colour in solution involving pre-filtration through sterilizing filters followed by the measurement of absorbance at several wavelengths. They did not perform a statistical correlation between visual and instrumental measurements. The South African national specification (SABS 49:1985) and the International specifications (BP:1980, BS 757:1975, GMIA:1986, USP:1995, etc.) for gelatine lack a method for the objective measurement of gelatine colour or clarity. It was also known that the Gelatine Manufacturers Institute of America (GMIA) have been considering a spectrophotometric method for clarity determination based on absorbance at 640 nm. While this approach might be acceptable if only pale pigskin gelatin were in the market place, it assumed that gelatin did not absorb light at 640nm and thus any observed "absorbance" must be due to light scatter due to imperfect clarity. With gelatins derived from bovine hide and other raw materials this assumption was obviously not true (see Fig. 2). Finally, since the incorporation of the Gelatin and Glue Research Association into the Leatherhead Research Association there has been no independent body researching problems like the measurement of gelatin colour, taking into account available technology and all the different gelatines manufactured world-wide. Hence, it was hoped that the broad range of gelatins covered in this study would demonstrate the possibility that agreement could be reached on a national and possibly an international standard for gelatine colour measurement.

This study investigates the problem of gelatine colour measurement and presents a simple and practical solution based on the use of the BYK-Gardner Color-view (reflectance) Spectrophotometer (BYK-Gardner GmbH, Lausitzer Str. 8, D-8193 Geretsried, Germany) for the objective measurement of gelatine solution colour which is in good agreement with the visual colour assessment method.


Commercial and laboratory gelatines (Type A from acid processed collagen and Type B from alkali processed collagen) from a number of manufacturers and from a range of sources, including pigskin, were used in the development of the methodology.

The preparation of gelatine solutions was the same as for the measurement of the Bloom gel strength (BS 757:1975). The 6.67% w/v sample solutions were held at 42 ± 2C prior to the measurement of absorbance, visual colour or turbidity.

The absorbance of gelatine solutions was scanned between 400 and 700 nm using a Beckman DU70 single beam spectrophotometer (Beckman Instruments, 2500 Harbour Boulevard, Fullerton, California, USA.) Distilled water was used as the blank. From the scans, the absorbance at any included wavelength could be obtained. The absorbance of duplicate samples of each gelatine were determined and the average of the two values was reported.

A Turbidimeter (ICM Turbidimeter. ICM. 163 S.W. Freeman, Hillsboro. OR 97123. USA) was used for the measurement of clarity/turbidity. The instrument was calibrated to read 40 NTU using a 40 NTU standard supplied by the manufacturer of the instrument. A portion of the gelatine solution (6.67% as used for the absorbance determination above) was used to fill (20 ml) the Turbidimeter cuvette and the turbidity was measured on the scale of 0 to 100 NTU.

Visual colour measurement was made by diluting 60 ml of the 6.67% gelatine solution to 100 ml with distilled water at 40C. This solution was compared to a 4% solution of a standard gelatine with an ascribed colour of 8 DCU, in 100 ml Nessler tubes. Solution was poured out of the darker tube until the colours were judged to be equal. The unknown gelatine was then ascribed a colour in conformity with Beer's law: VsCs = VuCu, where V = volume, C = colour value, s = standard and u = unknown. For example, if 70 ml of standard (8 DCU) matched 100 ml of unknown then the visual colour ascribed to an unknown was 70x8/100 = 5.6 DCU. If 40 ml of unknown matched 100 ml of the standard then the colour of the unknown was 8x100/40 = 20 DCU. For the accurate measurement of the visual colours of pale gelatines the standard was diluted to 2% concentration and 100 ml of standard had the ascribed colour value of 4 DCU. It should be noted that if differences in hue were observed every effort was made to obtain a colour match based on "lighter" or "darker" only which would be expected to equate to the "L" value of the tristimulus colour.

For colour measurement using the BYK-Gardner Color-view (reflectance) Spectrophotometer, the instrument was first standardised against the black and white tiles as prescribed in the instruction manual. The instrument was then set to measure the "standard" consisting of 50 ml of distilled water in the quartz sample cuvette provided (90 ml capacity) with the white tile covering the cuvette. Thereafter, 6.67% gelatine solutions (50 ml from a measuring cylinder) at 40C were measured using the same configuration. The instrument was programmed to measure each sample 3 times and to report the average colours on the L,a,b tristimulus scale as a comparison to the standard. "L" is a measure of darkness, "a" is a measure of red to greenness and "b" is a measure of yellow to blueness.

The data from the above determinations were entered into a Quattro spreadsheet (Borland International, 4585 Scotts Valley Drive, Scotts Valley. CA 95066. USA) for the calculation of correlation coefficients and linear regression equations. In the case of absorbance correlations, the absorbance of a colourless gelatine solution should be zero by definition, hence the regression coefficient was taken as zero. For three way partial linear regression analysis the appropriate data was entered into a Basic program which used the formulae of Spiegel (1961).


If one considered the visual perception of the colour of a solution, it was evident that both reflected and transmitted light reached the eye and based on this mixed stimulus an assessment of colour was made. Using this rationale, it was decided to attempt the measurement of colour using the BYK-Gardner Color-view (reflectance) Spectrophotometer with an arrangement that ensured that both reflected and transmitted light were received by the detector (Fig. 1)

Fig. 1. The BYK-Gardner Color-View spectrophotometer light-path using the
setup proposed for gelatine solution colour measurement. Molten solution
from the Bloom gel strength determination (50 ml) was measured into the quarts
cuvette provided with the instrument.

Furthermore, in this arrangement there was a large amount of scatter by the white tile and it became apparent that provided the scatter by the gelatine solution was small compared to that of the tile, the response of the instrument was proportional to gelatine colour. This resulted in an acceptable instrumental measurement of colour on a molten Bloom gel strength sample with a clarity of <80 NTU, there being no need for further sample preparation.

The absorbance spectra of gelatines over the visual range of wavelengths are shown in Fig. 2 from which it could be seen that darker gelatines have a greater absorbance at both 700 nm and 400 nm. The absorbance at 700 nm, bordering on the invisible infrared region of the spectrum, could be a source of error in the determination of colour based on 400 nm absorbance (A400) values.

From Fig. 2 it was concluded also, that the colour intensity of gelatine should be best related to the absorbance at 400 nm. The error due to scatter (possibly causing some of the 700 nm absorbance)

Fig. 2. The absorbance spectra of 6.67% aqueous solutions of four gelatines of good clarity, spanning the normal Davis Colour range.

Fig. 3. The correlation of the visual Davis Colour value of a range of gelatines and the Instrumental Davis Colour values obtained using linear regression and absorbance at 400 nm. uncorrected for turbidity; - - - - corrected for turbidity.

might be eliminated by use of a correction factor based on turbidity measurement which left an unknown effect due to 700 nm absorbance. Some typical data on gelatines with turbidities of <80 NTU is shown in Fig. 3, from which it was observed that the gelatine colour calculated from the regression Equation 1 had a very significant (>0.9995) correlation coefficient of 0.94. However, the actual differences between instrumental colour and visual colour exceeded the known error of the visual determination of 1.5 DCU, in some 33% of the determinations. From experience it was known that turbidity made the visual determination of colour difficult and this usually resulted in visual estimates being darker than expected. Hence, using partial multiple linear regression, the estimates of colour based on 400 nm absorbance, were corrected for turbidity (Equation 2). This had the result of improving the correlation coefficient marginally and improving the y-axis intercept from 2.2 to 1.0. However, the error of the instrumentally estimated colours were largely unaffected and still some 33% differed from the visual colour by more than 1.5 Davis Colour units (DCU).

The correlation of A400 absorbance and visual colour for 23 gelatine from various sources but with turbidities of less than 80 NTU gave the following linear regression equation:

DCU = 34.188 x (A400)...............................................(1)

Because, by definition, a visual colour of 0 must have an absorbance of 0 the regression equation was forced to a coefficient of 0. Hence the fact that the plots of the calculated and visual colours in Fig. 3 do not have a y-axis intercept of 0 indicates considerable bias initially caused by turbidity and then after correction, by the residual 700 nm absorbance.

Refinement of equation (1) to take into account light lost due to turbidity using partial multiple linear regression gave:

DCU = -0.914 -0.044 x NTU + 43.811 x (A400).........................(2)

Colour was the dependent variable 1, turbidity was the independent variable 2 and absorbance was the independent variable 3. The respective correlation coefficients were:

R12 = 0.695 R13 = 0.935 R23 = 0.799

In summary, Fig. 3 clearly depicted the problem of gelatine colour measurement and demonstrated that included turbidity was an inadequate solution. Furthermore, it was found that as the turbidity of the gelatines increased so the correlation coefficients decreased (not shown). Currently, because the majority of commercial edible gelatines have turbidities of less than 80 NTU this should not be an unacceptable restriction on the instrumental measurement of gelatine colour.

For the study using the BYK-Gardner Color-View spectrophotometer the normal spectrophotometric procedure of using the solvent as the standard or blank was adopted. This instrument presented colour in tristimulus L, a, b values as well as in terms of the DIN and ASTM Yellowness indices, but these indices were not found to give a better correlation with visual colour than the tristimulus L value.

Figures 4 to 6 show the correlation of visual colours of 67 gelatines (with turbidities of less than 80 NTU) with the L, a, b, tristimulus colour values based on water as the blank. (Included in the samples were production or experimental gelatines produced from bone, bovine hide, pigskin, horse hide, sheep skin, fish skin, and chrome tanned leather, including commercial gelatine blends. The gelatines were obtained from Leiner Davis Gelatine (South Africa) and manufacturers in Europe, Australia, New Zealand and North and South America).

Fig. 4. The Davis Colour value of 67 gelatines plotted against the tristimulus "L" value of 50 ml of the 6.67% molten Bloom gel strength sample using the setup in Figure 1 and a BYK-Gardner Color-View spectrophotometer.

Fig. 5. The Davis Colour value of 67 gelatines plotted against the tristimulus "a" value of 50 ml of the 6.67% molten Bloom gel strength sample using the setup in Figure 1 and a BYK-Gardner Color-View spectrophotometer.

Fig. 6. The Davis Colour value of 67 gelatines plotted against the tristimulus "b" value of 50 ml of the 6.67% molten Bloom gel strength sample using the setup in Figure 1 and a BYK-Gardner Color-View spectrophotometer.

It can be seen that the correlation between the negative L value (darkness) and visual colour was obviously linear and the correlation coefficient (r = -0.97) was markedly better than the correlations between visual colour and the "a" (+ = redness, - = greenness) or the "b" (+ = yellowness, - = blueness) components of the tristimulus colour values, as shown in Figures 5 and 6. Furthermore, these gelatines gave the following correlation coefficients between the visual colours and the instrumental values:

L = 0.97 a = 0.84 b = 0.89

DIN Yellowness = 0.91 ASTM Yellowness = 0.88

From Fig. 5 it appeared that the tristimulus "a" (red/green) component of gelatine colour had a second order correlation to visual colour with paler gelatines having a greener hue and darker gelatines a redder hue, however this phenomenon could be a function of the source of the gelatines, as all the darker gelatines were of South African origin.

From Fig. 6 the "b" value of the tristimulus colour appeared to indicate a continuously increasing yellowness with depth of colour.

Based upon the linear regression of visual colour and tristimulus "L" value the following equation was obtained:

DCU = - 2.79698 - 1.794101 (L value)................................(3)

with a correlation coefficient of 0.97. The resulting calculated Davis Colours had the same correlation coefficient (0.97) and a y-axis intercept of -0.4 DCU which indicated a small bias towards high estimates of colour. The standard error of the estimate of Davis Colour was 0.45 colour units. Finally, the "unacceptable" estimates of Davis Colour were reduced to 16%, six being higher and five being lower than the visual Davis Colour values.

Fig. 7. The day-to-day repeatability of the tristimulus "L" value of 50 ml of the 6.67% molten Bloom gel strength sample, using the setup in Fig. 1 and the BYK-Gardner Color-View Spectrophotometer.

The reproducibility of the measurement of tristimulus "L" value is shown in Fig. 7 for values of the same gelatine determined on different days. This data confirmed that the maximum difference between any two determinations of Davis Colour should not exceed 0.9 units which was a considerable (40%) improvement on the error of the visual colour value.


A rapid and reliable instrumental method for the measurement of gelatin colour in solution has been developed using a BYK-Gardner Color-View Spectrophotometer and a special arrangement of the apparatus as shown in Figure 1. The method is based upon the determination of the tristimulus "L" value of a gelatine solution obtained by simply melting the 6.67% solution used for the Bloom gel strength determination. Pale gelatines had an "L" value close to 0 and dark gelatines a value up to -15. These values were linearly correlated to colour values in conformity with Beer's law. The method was found to be almost free of interference (due to turbidity) as long as the turbidity of the 6.67% solution was less than 80 NTU. Furthermore, the method was suitable for routine laboratory use as it could be applied directly to a remelted Bloom gel strength samples. A colour scale based on L value x (-10) can be proposed.


BP:1980. British Pharmacopæia. London Her Majesty's Stationary Office. p.207.

BS 755:1975. British Standards Institution Specification for Gelatine. Pentonville Rd, London. UK.

Cole, C.G.B. (1995). Ph.D Thesis. University of Pretoria. Pretoria. RSA.

Hoffmann, P. (1985). Deutsche-Milchwirtschaft. 35, 269.

GMIA:1986. Gelatin Manufacturers Institute of America, Inc. Standard Methods for the Sampling and Testing of Gelatins. Suite 908, 271 Madison Ave, New York. NY 10016.

Ottenbacher, H. (1981). Suesswaren. 25, 24.

SABS 49:1985. Specification for Gelatine. South African Bureau of Standards. Pretoria. RSA.

Saunders, P.R. and Ward, A.G. (1953). Journal of the Science of Food and Agriculture. 4, 523.
Schreiber, R. (1977). CCB Review for Chocolate, Confectionery, and Bakery. 2, 14.

Siebert, J. (1992). Chemische Industrie. 115, 32.

Spiegel, M.R. Theory and problems of statistics. Schaum's Outline Series. Mc Graw-Hill. New York, St Louis, San Francisco, Sydney. 1961 p. 269.

USP:1995. The United States Pharmacopeia The National Formulary. United States Pharmacopeial Convention, Inc. 12601 Twinbrook Parkway, Rockville, MD 20852. p. 2247.

Veis, A. The macromolecular chemistry of gelatin. Academic Press - New York and London. 1964. p. 64.


Submitted in partial fulfilment of the requirements for the degree of Ph.D (Food Science), University of Pretoria.

Thanks to Leiner Davis Gelatin (South Africa) for their support.