AOAC Methods in Codex STAN 234 (Preliminary Methods Review)

BARBANO ET AL.: J. ASSOC. OFF. ANAL. <.'HEM (VOL. 7l. NO. 6, 19qo) 855 r.a,1, 1. Percent protein detennlned by KJefdahl method on raw milk eamples, flrat lrlal of collaborative 1tudy (dlgNted 1 h after clearlng) L.abOralory

Material

Mlllwlal

B 3.457 3.419 3.408 3.375 3.097 3.10.9 3.618 3.612 3.076 3.108 3.133 3.148 3.202 3.268 3.338 3.367 3.273 3.279

A 3.488 3.505 3.477b 3.387b 3.101 3.084 3.680 3.618 3.068 3.116 3.138b 3.197b 3.268 3.256 3.411 3.385 3.264b 3.298b

C

0

E

F

H

mean

G

K•

J

3.439 3.439 3.382 3.385 3.094 3.082 3.612 3.610 3.082 3.096 3.132 3.132 3.262 3.265 3.363 3.371 3.266 3.271

3.431 3.414 3.393 3.408 3.085 3.085 3.554 3.593 3.071 3.118 3.136 3.142 3.246 3.27'4 3.375 3.314 3.273 3.261

3.429 3.431 3.385 3.380 3.070 3.078 3.599 3.608 3.081 3.077 3.141 3.128 3.235 3.243 3.359 3.370 3.258 3.239

3.424 3.424 3.396 3.374 3.054 3.068 3 ,582 3.580 2.912" 3.080° 3.124

3.472 3.439 3.395 3.408 3.086 3.087 3.620 3.698 3.099 3.067 3.139 3.138 3.288 3.271 3.366 3.351 3.277 3.233

3.467 3.462 3.421 3.411 3.108 3.100 3.634 3.637 3.109 3.110 3.153 3.152 3.276 3.270 3.399 3.393

3.422 3.426 3.384 3.392 3.073 3.068 3.584 3.594 3.061 3.076 3.139 3.121 3.228 3.263 3.345 3.349 3.283 3.230

3.466 3.457 3.406 3.418 3.092 S.110 3.625 3.637 3.080 3.147 3.158 3.165 3.264 3.281 3.583d 3.397d

3.536 3.494 3.448 3.448 3.100 3.140 3.670 3.640 3.132 3.138 3.208 3.160 3.268 3.313 3.424 3.436 3.328

3.446

2

3.396

3

3.086

4

3.610

5

3.091

6

3.140

3,131 3.235 3.217 3.372 3.359 3.242 3.251

7

3.259

8

3.367

9

3.281 3.275

3.298 3.290

3.266

3.3i7

n Invalid csata (laboratory did not use class A buret). b tnvalld data (char ring)

•Statistical outlier (Grubbs test). d Stallstlcal outlier (Cochran test).

and distill under same conditions as for a milk sample. Re- 4'0Veries shall be at least 98%. rt CalculalloM SeeG, Q. Rt,peatab/llty andReprodur:lblllly Val11es For method performance parameters obtained in collabo– rative study of this method, r value = 0.038 and R value = 0.049.

were 3 invalid data in laboratory A; materials 2, 6, and 9. Low test results on these samples were caused by a char ring of undigested material on the block digestor tubes after di· gcstion (reported by analyst). Data from laboratory K were found to be invalid; this laboratory did not use a class A buret for titration (this was discovered after trial 2 of the collab;.>rative study). The buret that was used was found to deliver less acid than indicated, which may explain why this laborat,ory's results were consis– tently higher than results from the other laboratories. There were 2 statistical outliers. Laboratory F material 5 was an outlier by the Grubbs test (the value was significantly lower than values from the other laboratories). The test value for laboratory J material 8 was a statistical outlier by the Cochran test. The mean test results are reported in Table l by material. Blank values from laboratory to laboratory ranged from 0.05 to 0.20 mL. This range indicates that each laboratory must conduct blank determinations to correct for its own conditions. If a laboratory obtains an unusual blank value on a daily analysis, the cause of the unusual value should be determined. The AOAC statistical parameters for each sample materi– al are shown in Table 2. The mean AOAC statistical values are also stated. The method showed very good repeatability and reproducibility. The s, and r vaEues reported in the IDF provisional method (16) were 0.016 and 0.045, respectively, which are similar to those reported here. When the AOAC statistical parameters from Table 2 were evaluated as a func– tion of protein concentration for all data, linear regression indicated no significant influence of protein concentration on any of the AOAC statistical parameters (Table 3). Traditional and block digestor data were separated and statistical analysis was done separately for traditional equip– ment and block equipment to compare their performance (Table 3). The 2 different types of equipment performed similarly when used with the method in this study. Linear

Ref.: JAOAC 73, November/December issue ( 1990),

Results and Discussion

V•rltlcallon of MIik Splntl"fl Accuracy and MIik Quality The range in percent fat of the milks (as measured with a Multispec Dairylab 2) was never greater than 0.02%, for both colfaborative study trials, when the individual sample materials were checked for splitting accuracy by determining the fat content of milk in the first, middle, and last bag of milk poured. The average range across the first, middle, and last bag of milk was 0.01% fat, which indicated that the Jamples were uniformly split prior to shipment to collaborat– ing laboratories. All of the 18 milks {trial I end trial 2) had a somatic cell count of ~6.4 X l0 5 cells/mL, with a mean somatic cell count of 4.7 X 10 5 cells/ml. Flesults of Co/laborat/1111 Trial 1 All of the data collected for total nitrogen determination by the Kjeldahl method from traditional and block equip– ment for the first collaborative study are reported in Table I . Blind duplicates have been paired for each sample material. Laboratories A, C, D, B, G, and I used block digestor plus steam distillation equipment. The remaining laboratories (B, F, H, J, and K) used traditional Kjeldahl equipment. There

Codex Trial Method Review