Sie sind auf Seite 1von 3

ISO 9001


Degussa AG • Feed Additives • D-63403 Hanau amino acids and more.

Dumas or Kjeldahl
Is there an influence on crude protein measurement?

Interest has been expressed in understanding the differences between the Dumas nitrogen deter-
mination and the Kjeldahl method. The Dumas or Combustion Nitrogen Analysis method (CNA)
was developed in 1831, thus it predates the Kjeldahl method by over 50 years.

It should be mentioned that neither of these methods measures protein. They measure nitrogen,
from which crude protein is calculated. The nitrogen content in percent is multiplied by 6.25 for
typical feed ingredients and mixed feeds to come up with the crude protein level. This number is
derived from the fact that the average proteins contain 16% nitrogen, thus 6.25 is the reciprocal of
0.16 or 16%.

Other factors are used for special products. For example, when milk proteins are measured, a
factor of 6.38 is used due to the fact that milk proteins contain about 15.66% nitrogen rather than
16% that is found in plant and animal proteins. This is also why this measure is called crude pro-
tein, not protein or true protein. It is a calculated value, not a measured value. We measure nitro-
gen and calculate crude protein. Crude protein also contains Non-Protein-Nitrogen (NPN) sources
additionally to pure proteins (see below).

Another consequence of this is that the sum of the analyzed amino acids is not 100%, compared
to crude protein, but commonly 10% or more below. When calculating the net protein, based on
accurate analyses of all amino acids and comparing it to the nitrogen content, a smaller conversion
factor than 6.25 is found. P. Salo-Väänänen and P. Koivistoinen (1996) investigated food samples
and proposed factors of 4.94 (fish products), 5.17 (meat products), 5.40 (cereals), 5.51 (processed
foods) and only 5.94 (milk products). Thus according to these authors for individual raw materials,
the protein contains 17 to 20% nitrogen rather than the above 16%. A problem is also that the
nitrogen of a sample is not only coming from amino acids, but also from several non-protein-nitro-
gen (NPN) sources (urea, nucleic acids, alkaloides, amino sugars, lecithine, betaine, choline etc.).

These days the Combustion Nitrogen Analysis (CNA) method is becoming the method of choice
for many laboratories due to the fact that:

1. It extracts practically all nitrogen present in the sample.

2. The Kjeldahl method requires the use of concentrated sulfuric acid, hydrogen peroxide and
50 % sodium hydroxide, all of which require special care from the laboratory personnel.

3. Kjeldahl uses various highly toxic heavy metal catalysts, such as copper, mercury or sele-
nium, which require expensive disposal methods.

4. The CNA does not need these materials in addition to the fact that it is a much faster method
of analysis. A typical CNA takes only about 5 minutes, while a Kjeldahl takes longer than 2
hours for chemical digestion, distillation of ammonia and titration with hydrochloric acid.
ISO 9001
Degussa AG • Feed Additives • D-63403 Hanau amino acids and more.

These two methods do not generate identical data. The CNA converts more nitrogen than the
Kjeldahl method does. This is caused by better recovery of amino-nitrogen, but also due to the
fact, that even nitrates are transformed to nitrogen gas. In the enclosed Table comparison data
from our lab, as well as from several international ring tests (mean values) are shown. On average
CNA determines nitrogen about 1.7% relatively higher than the Kjeldahl method. This results
in absolute crude protein values measured by the Dumas method that are about 0.15 to 0.25%
points higher for wheat and 0.5 to 0.7% points higher for soybean meal.

The Degussa AminoNIR® crude protein prediction equations were developed based on the CNA
method. It is not unexpected that the Degussa calibration equation for crude protein would predict
a higher crude protein than the same sample run via Kjeldahl, due to the fact that the calibration
was developed using the CNA method, which extracts more nitrogen, thus increasing the
calculated crude protein content. This difference becomes more obvious at higher ingredient crude
protein levels. Thus the difference between the crude protein content of a high protein ingredient,
such as meat and bone meal or feather meal, found by the Kjeldahl and by the Dumas will be
greater. At low crude protein levels, such as in corn or sorghum, the difference between the
Kjeldahl and the Dumas method will not be that obvious.

However one has to consider, that the standard error of cross validation SECV for CP (NIRS) is
typically 1.7 – 2.3% relative to the mean crude protein content of the calibration population. Based
on a 95% confidence level this results in a relative range of 3.4 – 4.6% around the NIRS crude
protein prediction, in which the wet chemistry value would be expected. As this prediction error is
clearly higher than the typical difference between Kjeldahl and Dumas, the influence of the refer-
ence methods is negligible.

However all these effects in crude protein determination, have no influence on the accuracy or
precision of the Degussa NIRS amino acid prediction equations. These equations were devel-
oped based on individual amino acid, wet chemistry analyses. Each amino acid calibration equa-
tion was developed independently of other amino acids and independently of the crude protein
prediction equation. Thus it is perfectly appropriate to use the amino acid values from the Degussa
AminoNIR calibrations for formulation purposes, even if the local laboratory running Kjeldahl finds
somewhat lower crude protein values than that predicted by the Degussa NIRS calibrations.


- P. Salo-Väänänen & P. Koivistoinen, Food Chemistry, Vol. 57, 27-31 (1996)

- Unpublished ring test results from years 1999 - 2005
ISO 9001
Degussa AG • Feed Additives • D-63403 Hanau amino acids and more.

Comparison of results for the nitrogen determination with the Kjeldahl method and the
Dumas combustion method in animal feed

Sample Ring Test No. of Labs CP Kjeldahl CP Dumas Relative difference

Organisation (Kjel/Dumas) mean (%) mean (%) (%)
Rice hulls USA (6/98) 2.39 2.43 1.67
Soybean meal USA (3/91) 47.81 48.60 1.65
Soybean meal USA (7/93) 50.35 51.05 1.39
Fish meal USA (4/91) 62.04 63.11 1.72
MBM/Hulls mix USA (3/90) 31.44 31.87 1.37
Feed for Calfs USA (5/114) 19.67 20.11 2.24
Feed for Broiler USA (5/108) 27.45 27.89 1.60
Feed for Pigs USA (4/106) 21.21 21.67 2.17
Feed for Chicken USA (5/111) 18.27 18.55 1.53
Feed for Piglets USA (4/107) 21.46 21.69 1.07
Feed for Dogs USA (5/109) 27.24 27.35 0.40
Feed for Cattles USA (6/98) 42.05 43.33 3.04
Feed for Chicken USA (7/111) 17.64 18.05 2.32
Feed for Lambs USA (6/111) 17.76 18.08 1.80
Feed for Piglets USA (7/113) 16.44 16.79 2.13
Feed for Piglets USA (7/111) 19.79 20.33 2.73
Feed for Calfs USA (4/113) 15.51 15.80 1.87
Feed for Lambs USA (4/108) 21.81 22.15 1.56
Feed for Cattles USA (5/116) 22.44 22.94 2.23
Feed for Piglets USA (5/102) 23.01 23.14 0.56
Feed for Piglets USA (4/109) 16.24 16.40 0.99
Feed for Piglets Germany (44/17) 21.10 21.58 2.27
Mixed Feed Germany (43/17) 20.11 20.49 1.89
Feed for Turkeys Germany (20/5) 22.75 23.19 1.93
Mixed Feed Germany (20/5) 20.39 20.72 1.62
Dairy Feed Germany (15/13) 22.84 23.24 1.75
Feed for Piglets Germany (15/14) 21.09 21.52 2.04
Dairy Feed Germany (14/5) 21.59 22.07 2.22
Feed for Piglets Germany (15/5) 21.18 21.63 2.12
Dairy Feed Germany (15/14) 22.08 22.56 2.17
Feed for Turkeys Germany (15/14) 22.40 22.78 1.70
Feed for Piglets Germany (17/14) 21.07 21.43 1.71
Dairy Feed Germany (17/13) 20.42 20.84 2.06
Alfalfa meal AminoLab 16.71 16.83 0.67
Fish meal AminoLab 77.05 77.47 0.54
Fish meal AminoLab 66.42 67.03 0.92
Corn silage AminoLab 9.28 9.58 3.30
Corn silage AminoLab 9.98 10.23 2.51
Rapeseed meal AminoLab 31.79 31.83 0.12
Rapeseed meal AminoLab 34.39 34.54 0.44
Rapeseed meal AminoLab 34.43 35.14 2.05
Soybean meal AminoLab 51.45 52.48 1.99
Soybean meal AminoLab 44.73 45.39 1.47
Soybean meal AminoLab 44.17 44.26 0.20
Sunflower meal AminoLab 27.24 27.71 1.70
Triticale AminoLab 12.02 12.08 0.52

Relative Difference Kjeldahl method - Dumas method Mean 1.7