Infrared Milk Analyzer Research Articles

Accuracy and Repeatability of Milk Total Solids Analysis by Microwave Moisture Oven

A microwave moisture tester (CEM AVC80) was evaluated as an alternative to the official AOAC (Association of Official Analytical Chemists) oven drying method for calibration of the solids-non-fat function of Multispec infra-red (IR) milk analyzers. Reference samples used to calibrate the IR instruments were analyzed by both AOAC and CEM procedures during a period from October to December, 1987. Chemical solids-non-fat (SNF) values were determined as AOAC or CEM total solids less Mojonnier fat. SNF values based on each drying method were independently used to calibrate the SNF function of the IR instruments and compared using mean difference values (estimates less chemical values). Residual plots showed uniform distribution of residuals with respect to SNF levels indicating that the calibration models fit both CEM and AOAC data equally well. The CEM produced mean total solids values 0.09% higher than AOAC values. Two separate repeatability trials showed that a 1988 model of the CEM AVC-80 was more repeatable than the 1987 model (standard deviation of duplicate differences was 0.020 versus 0.029) but both were less repeatable that the AOAC method (standard deviation of differences was 0.011).

Dry Calibration Milks for Infrared Milk Analyzers

Abstract A study has been carried out to determine the practicality of formulating dry, stable calibration milk powders for calibration of infrared milk analyzers for routine quality control purposes in commercial dairies. The powders are based on dairy Ingredients and are reconstituted to produce calibration milks that cover a wide range of fat, protein, and lactose composition combinations. The butterfat component included emulsifiers to aid in reconstituting the powders and to stabilize resulting emulsions. Reconstitute was carried out on a weight/weight basis using a top-loading balance and mixing with a simple hand-held, electric household mixer. The reconstitutlon procedure produced stable emulsions and was shown to be repeatable to within the accuracy specifications of Infrared instrumentation. The calibration results obtained with the powders relative to reference calibration milks purchased from the Central Milk Testing Laboratory (CMTL) In Guelph, Ontario, were excellent, and generally met AOAC specifications for Instrument calibration. In addition, the reconstituted milks had excellent emulsion stability to prolonged holding at 40°C and produced consistent results when analyzed over a 21-day period, showing no signs of spoilage when preserved with potassium dlchromate and refrigerated when not In use. The powders were deemed adequate for calibration of Infrared milk analyzers used for routine, In-plant, quality control purposes and would be a convenient alternative for CMTL calibration milks shipped by courier.

Infrared Milk Analysis — Challenges for the Future

Infrared milk analyzers are the only practical method of economically testing large numbers of milk samples for fat, protein, and SNF content in producer payment and dairy herd improvement programs. The major challenge for successful use of infrared analyzers is correctly establishing and maintaining accurate calibration. To achieve an accurate calibration, an infrared analyzer needs to be checked periodically to verify that zero drift, repeatability, purging efficiency, linearity, and homogenization efficiency meet minimum instrument performance standards. Once these performance criteria have been satisfied, the primary slope for each channel can be established and maintained. Next, intercorrection factors for each component need to be determined. Two different approaches are used to determine intercorrection factors for an instrument: characterization with reference solutions or multiple regression of uncorrected instrument data versus chemical data on large numbers of milk samples. The comparative ruggedness of these two approaches has never been collaboratively studied. Finally, secondary slope and bias for day-to-day adjustment of calibration are established. In instruments with both fat A and B filters, the ratio of these signals used for fat determination needs to be set. Various approaches for combining the A and B filter signals have been proposed, but the relative ruggedness of different approaches has not been collaboratively studied. In the future, uniform and highly accurate procedures need to be developed for infrared analyzer calibration based on collaborative studies.

Proximate Analysis of Fish Tissue by Mid-Infrared Transmission Spectroscopy

Mid-infrared spectroscopy was assessed as a means for rapidly determining the fat, protein, and moisture content of fish tissues. For fat and protein, a sample preparation protocol was developed for the conversion of fish tissue into a milk-like emulsion for analysis by a Multispec MK I infrared milk analyser. Fish (cod, tuna, and mackerel) were minced, dissolved, and emulsified in 0.2 M NaOH using a polytron and analysed for fat and protein using the conventional fat and protein wavelengths used for milk analysis. Instrument calibration for fat and protein was based on the Mojonnier and Kjeldahl procedures, respectively. Moisture was determined separately by extracting the water from the minced tissue using absolute methanol and measuring the absorbance of the resulting solution using a 5.56/6.02-μm filter pair. The instrument was calibrated using standardized water–methanol mixtures and the results compared with the standard vacuum oven method. Rapid analyses of fat and protein were possible giving individual results to within 1% of the standard chemical determinations. Moisture analysis, also worked well, being more accurate and rapid than the vacuum oven procedure. The limiting factor in its use for proximate analysis appears to be a combination of sample compositional variation and/or the general reproducibility of the standard methods. We conclude that the analysis of fish by mid-infrared transmission spectroscopy could serve as a means of rapidly carrying out the proximate analysis of fish and fish products for quality control purposes.

Factors Affecting Differences in Milk Fat Test Obtained by Babcock, Rose-Gottlieb, and Infrared Methods and in Protein Test from Infrared Milk Analysis

In a collaborative study between personnel of the Canadian dairy cattle Record of Performance Services, university laboratory, and central milk testing laboratory of Quebec Dairy Herd Analysis Service, differences in fat tests by Babcock, Rose-Gottlieb, and infrared analyses (Milko-Scan) were studied in individual milk samples from Ayrshire, Holstein, and Jersey herds for 1 yr. On-farm Babcock for fat gave results that were lower by .075, .058, and .016% than laboratory Babcock tests for Ayrshire, Holstein, and Jersey milks. Fat tests by Rose-Gottlieb and Milko-Scan on day of sampling were intermediate between the on-farm and laboratory Babcocks for Ayrshire and Holstein. For Jersey milk, using Rose-Gottlieb as reference method, fat would be overestimated by .021, .037, and .082% when compared with on-farm Babcock, laboratory Babcock, and Milko-Scan results. Lower fat tests were obtained by infrared for 3-d-old samples transported under varying ambient temperature conditions; however, these were not due to error of the instrumental method per se but rather to conditions of milk samples which may have caused the milk to undergo lipolysis. Magnitude of the d 3 fat depression, which varied from nonsignificance to .21%, was accompanied by an increase in protein reading and was also dependent on breed of cow, stage of lactation, and month of the year.

Rapid Method for Determination of Milk Casein Content by Infrared Analysis

Preliminary work was to develop a method for determination of milk casein content using commercially available infrared milk testing equipment. Fresh whole milk samples were split into two portions. One portion was tested for total protein with an infrared milk analyzer. The other portion was adjusted to pH 4.6 with a specific volume and concentration of phosphoric acid. Milk casein precipitated at pH 4.6 and was removed by filtration. Next, the noncasein protein filtrate was tested for protein content using an infrared milk analyzer. The difference between the percent protein of the original milk sample and the percent protein of the noncasein protein filtrate equals the percent milk casein. The infrared casein determination for 36 different individual herd milk samples was not significantly different from results obtained by the official International Dairy Federation casein testing method. The infrared method is faster than the chemical method and could be done with infrared milk testing equipment that is commonly available in many cheese manufacturing plants. Repeatability of the infrared method was comparable to the repeatability of the chemical method.

A Study of the Stability of Record of Performance Milk Samples for Infrared Milk Analysis

The effect of switching individual cow Record of Performance milk samples from on-farm Babcock analytical system to centralized infrared analysis was investigated. The effects of lipolytic activity on the signals of the fat (5.73/5.58 and 3.48/3.56μm) and protein (6.46/6.68μm) wavelengths of filter instruments showed that lipolysis decreased the infrared fat signal and concurrently increased the protein signal. The chain length (3.48/3.56μm) signal was unaffected by lipolysis due to the concordant movement of the reference wavelength baseline with the changes in the sample wavelength signal. Storage trials (3 d) of preserved milk samples indicated no direct relation between fat signal depression and chemically determined free fatty acid levels. Microbial growth and agitation also contributed to changes observed in samples studied under controlled conditions. Based on concurrent infrared, Babcock, and Mojonnier analyses of approximately 900 samples on d 1 and 3, there was a consistent but marginal drop in the carbonyl (5.73/5.56μm) fat signal and an increase in the amide (6.46/6.6μm) protein signal, but these changes were of limited significance. A study of 3-d and older samples indicated that significant levels of acid were produced, indicative of microbial growth. A comparison of Babcock results carried out on-farm, Mojonnier analyses in the laboratory, and their corresponding infrared results did not show statistically significant differences. Other than the inherent variability present in the individual cow samples, there were minor quality changes occuring under the present sample handling system due to a combination of lipolysis, microbial growth, and acid production. Refrigeration of Record of Performance samples during transport and at the central laboratory is recommended to minimize these changes.

赤外線牛乳分析器による補正値を用いる個体乳のTMS率算出

生乳の全乳固形分(TMS)率は,生乳の取引価格の算定,乳牛の改良,飼養管理などに関連する重要な測定項目である.個体乳のTMS率を赤外線牛乳分析器を用いて得た測定値から算出する際の問題点を検討した.千葉県安房地域(等間隔搾乳),農林水産省畜産試験場(不等間隔搾乳)で生産された個体乳試料について,朝乳•夕乳ごとに,脂肪•蛋白質•乳糖率をミルコスキャン203B(FossElectric)で,TMS率をTMSチェッカK375A(安立電気)で測定した.このTMS率と脂肪•蛋白質•乳糖(無水物換算)率の総和との差(α値)の変動について調べた.その結果,α値の平均値は他成分と異なり,搾乳間隔•季節•産次•泌乳期による影響をほとんど受けず,ほぼ一定の値をとった:安房=1.07±0.108,1-07±0.111,畜試=1.03±0.114,0.99±0.126(朝乳•夕乳の順, Mean±S. D.).個体乳の場合,α値のバラツキが大きいので,α値の平均値を補正値として計算して得たTMS率は目安程度の扱いとすべきだが,α値の平均値の安定性は注目に価する.

Laboratory Experiments with a New Infrared (IR) Milk Analyzer, the Milko-Scan 605

Abstract A new single-cell, dual-wavelength infrared milk analyzer, the Milko-Scan 605, has been introduced. The instrument is based on the same principle as the Milko-Scan 100 series. Various parts, however, have been modified (e.g., the detector, pump unit, and homogenizer), and the instruments has been supplied with a micro-processor. A number of critical points can be monitored and various possibilities for calibration and calibration control schemes can be preprogrammed. During the three-month trial, a traditional laboratory evaluation was performed at the Swedish Dairies Association. Performance characteristics such as repeatability, carry-over effects for different sample volumes, sample temperature effects, and so on were studied. Accuracy was calculated for fat (5.7 μm and 3.5 μm wavelengths), protein, and lactose determinations on bulk milk samples as well as samples from individual cows. The relative accuracies (coefficient of variation, CV) were estimated at 2.3% for fat (5.7 μm), 1.4% for fat (3.5 μ), 1.4% for protein and 1.3% for lactose. Interfering factors such as refractive index, free fatty acids, citric acid, and non-protein nitrogen were measured and statistically evaluated.

The mojonnier as a reference method for the infrared analysis of fat in meat products

The Mojonnier method was compared with the conventional Soxhlet method for the determination of fat in 7 different meat products to assess its use as a standard reference method for calibration of commercial quantitative infrared transmission analyzers (e.g., Multispec M, Milkoscan 300 or 104). Results for the meat samples obtained by the Mojonnier method did not differ significantly from those obtained by the Soxhlet method. In addition, the Mojonnier method was less time-consuming and more precise than the Soxhlet; therefore, it can be used as a standard reference procedure for the calibration and assessment of infrared milk analyzers in their potential application to the rapid determination of fat in meat and meat products.

Mojonnier Method as Reference for Infrared Determination of Fat in Meat Products

Abstract The Mojonnier method was compared with the conventional Soxhlet method for the determination of fat in 7 different meat products to assess its use as a standard reference method for calibration of commercial quantitative infrared transmission analyzers (e.g., Multispec M, Milkoscan 300 or 104). Results for the meat samples obtained by the Mojonnier method did not differ significantly from those obtained by the Soxhlet method. In addition, the Mojonnier method was less time-consuming and more precise than the Soxhlet; therefore, it can be used as a standard reference procedure for the calibration and assessment of infrared milk analyzers in their potential application to the rapid determination of fat in meat and meat products.

A bacteriostatic mixture for milk samples and its effect on bacteriological, cytological and chemical compositional analysis.

Of the bacteriostats tested, that comprising boric acid, glycerol, potassium sorbate and nystatin was the most suitable in preventing the multiplication of bacteria, yeasts and moulds in milk whilst causing least change in the Direct Epifluorescent Filter Technique (DEFT) count of bacteria and somatic cells. The DEFT and plate count of bacteria and the DEFT count of somatic cells were reduced after preservation and storage. The relationships between the plate count of fresh milk samples and the DEFT count and plate count of the same samples after preservation and storage for 3 d at room temperature (20–22°C) had correlation coefficients (r) of 0.80 and 0.85, respectively. The relationship between the Coulter count of somatic cells in fresh milk and the DEFT count of somatic cells in stored preserved milk had a correlation coefficient of 0.90. The addition of the bacteriostatic mixture to milk affected the determination of protein by the Milkoscan and lactose by the Infra Red Milk Analyser. It did not affect the estimation of the other milk components (fat, protein or lactose) by either instrument.

Influence of Dairy Rations with Whole Cottonseed or Extruded Soybeans on Accuracy of Analyses for Milk Components

To determine whether dairy rations influence the accuracy of component analyses, fat, protein, lactose, and total solids were measured in milk produced by Holstein cows fed a basal ration or a ration containing either 7% extruded soybeans, 15% whole cottonseed, or 30% whole cottonseed for the first 15 wk postpartum. Samples from individual milkings were collected in the 14th and 15th wk from 12 or 13 cows in each group. Reference methods were Mojonnier for fat, Udy dye-binding for protein, polarimetry for lactose, and forced-air oven for total solids. The methods compared to reference methods were Milkotester and Babcock for fat and Infrared Milk Analyzer for fat, protein, lactose, and solids-not-fat. Fatty acid composition was determined by gas chromatography. For cows fed 30% whole cottonseed, Infrared Milk Analyzer fat measurements averaged .14% lower than by Mojonnier. For cows fed extruded soybeans or whole cottonseed, Milkotester fat measurements were higher than by Mojonnier, and Infrared Milk Analyzer protein measurements were higher than by Udy dye-binding. For cows fed whole cottonseed, Infrared Milk Analyzer solids-not-fat measurements were higher than by forced air oven. For cows fed extruded soybeans or whole cottonseed, differences between Babcock and Mojonnier fat measurements and between Infrared Milk Analyzer and polarimetric lactose measurements were not affected. For cows fed whole cottonseed, the mean molecular weight of the fatty aicds of milk fat was increased.

Evaluation of Milkoscan 104 Infrared Milk Analyzer

Abstract The Milkoscan 104, a single cell, dual wavelength infrared milk analyzer, was assessed to see if it would meet the newly set AOAC specifications for accuracy and reproducibility required for official sanctioning of the instrument for milk component analysis. This assessment was carried out using herd, individual cow, commercial, and composite random milks, and produced mean differences and standard deviations of difference of <0.02% and ±0.02% for reproducibility and <0.05% and ±0.06% for accuracy. Results outside these values were encountered in the individual cow milks, for which the instrument was not specifically calibrated. The average molecular weight of the milkfat was an important variable because 2 anomalous milks were present in the calibration milks, thus restricting the overall accuracy obtainable. Linear regression analysis of chemical vs instrumental results showed good linearity and provided standard errors of the estimate of the same magnitude as the standard deviation of the differences. A separate assessment of the instrument for within- and between-day variation showed no practical significant differences. Based on the overall results, the Milkoscan 104 was judged to be capable of meeting the AOAC specifications set for fat, protein, and lactose analysis.

Performance Specifications for Infrared Milk Analysis

Maximum limits of 0.02% are recommended for precision of IR analysis for fat, protein, and lactose, and 0.04% for total solids. Recommended maximum systematic errors are 0.06% for fat, protein, and lactose, and 0.12% for total solids, when instrument results are compared with results of specified AOAC methods. Recommended maximum mean differences between instrument and standard results are based on 95% confidence limits for the mean of 8 samples; precision and systematic errors are calculated as the standard deviation of difference between duplicates for 8 samples, and systematic errors are calculated as the standard deviation of difference between instrument and reference results for 8 samples. Automated sections for milk of Official Methods of Analysis were rewritten to eliminate references to individual instruments, to eliminate possible differences between calibration procedures for different instruments, and to stress important instrument operational characteristics which can have significant effects on instrument performance.

Calibration and Standardization of the Infrared Milk Analyzer. The California Experience

Calibration and Standardization of the Infrared Milk Analyzer. The California Experience

Variation in Estimates of Milk Fat, Protein, and Lactose Content Asssociated with Various Bulk Milk Sampling Programs1,2

Three milk samples were taken from each bulk shipment from 26 herds for 13 mo; four samples were taken from each bulk shipment from 22 herds for a month. All herds were located in the Fraser Valley in British Columbia. Some samples (5656) were analyzed fresh with duplicate determinations on 955; the remaining 9842 were used to form herd composite samples (both 1-wk and 2-wk collection periods per composite). All samples were analyzed for milk fat, protein, and lactose by Infrared Milk Analyzers. These data were used to estimate the variation associated with various sampling and testing programs for bulk milk. The estimate of monthly herd mean milk constitute percentage from four simple random samples or three stratified random samples is expected to be at least as precise as the estimate from two 2-wk composite samples. Two-week composites underestimated percent milk fat by .045±.004 and overestimated percent protein and lactose by .023±.003 and .010±.002 compared to fresh samples. Thus, random sampling is expected to be cheaper and more accurate than composite sampling. Milk marketing areas considering changing from composite to random sampling may find useful the formula for calculating the number of random samples required to maintain the precision of composite sampling.

Influence of Feeding Protected Tallow on Accuracy of Analyses by Infrared Milk Analyzer

To determine whether feeding cows protected-tallow supplement influences reliability of analyses of milk composition by the infrared method, we measured fat, protein, and lactose in milk produced by three groups of seven Holstein cows fed supplement at 0 (basal), 15% (medium), and 30% (high) of their ration for the first 15 wk postpartum. Samples from individual milkings were collected every 3 wk from cows in each group. Reference methods for the Infrared Milk Analyzer were Babcock test for fat, Udy dye-binding for protein, and a polarimetric method for lactose. Fatty acid composition was determined by gas chromatography. The differences, reference method minus infrared, for cows in the basal, medium, and high groups were: for fat −.01, .29, and .30%; for protein, .08, −.02 and .03%; and for lactose .05, .25, and .21%. For cows in the medium and high groups, differences for fat were correlated with the mean molecular weight of the triglyceride fatty acids. Cows in all three groups were switched over to a common ration for wk 16 to 20, and the sampling schedule was modified so that two samples were collected from each cow at a 2-wk interval. Differences on this ration for cows in the basal, medium, and high groups were: fat −.05, .03, and .01%; for protein .03, .05, and .06%; and for lactose .12, .15, and .10%. Differences for fat and lactose when the cows were fed protected-tallow supplement were attributable largely to feeding the supplement and the resulting change in fatty acid composition of the milk fat. To avoid errors in measuring fat and lactose in milk from cows fed protected-tallow supplement by infrared analysis, the instrument must be calibrated specially.

Monthly Variations in Gross Composition of California Herd Milks

Data on the gross composition of herd milks were collected over 14 mo. Fifteen herds (five Jersey, four Guernsey, five Holstein, and one Shorthorn) from California's San Joaquin Valley were studied. An Infrared Milk Analyzer was used to determine fat, protein, lactose, and solids-not-fat in samples collected daily from farm tanks. Monthly variations were parallel for fat and protein concentrations in the milk from all breeds, winter milk containing more than summer milk. Monthly variations for lactose concentration were not as great as for fat and protein, with winter milk containing less than summer milk. Annual means for each component by breed did not differ from Illinois data collected 30 yr ago over 12 mo except for fat in Guernsey milk, which was .5% lower. Monthly variations in gross composition was the same in California as in Illinois. However, there appear to be differences from gross composition in 12 mo studies in South Dakota and North Carolina and for a shorter study in Pennsylvania.

Determining Fat, Protein, and Lactose in Milk from Cows Fed Formaldehyde Protected Sunflower-Soybean Supplement

To determine whether feeding cows protected lipid supplement influences reliability of analyses of milk composition, we measured fat, protein, and lactose in milk produced by two groups of six Holstein cows, one fed a conventional hay-concentrate ration and one a ration containing protected sunflower-soybean supplement. Analyses were of 36 samples from each group collected biweekly over 12 wk. The three constituents were measured by an Infrared Milk Analyzer. Reference methods were Babcock test for fat, Udy dye-binding for protein, and a polarimetric method for lactose. Fatty acid composition was determined by gas chromatography, and three nitrogen fractions (total, noncasein, and nonprotein) were measured by Technicon Auto Analyzer. The differences, reference method minus infrared, for cows fed the conventional and protected lipid rations, were for fat .09 and .46%, for protein −.02 and −.08%, and for lactose .06 and .23%. Thus, agreement between infrared and reference methods was closer for cows fed conventional rations than for those fed protected lipid supplement. For cows fed the supplement, the low fat by infrared was attributed to an increase in the mean molecular weight of the fatty acids. Changes in fatty acid composition of milkfat when cows are fed protected sunflower-soybean supplement causes errors in estimation of fat by infrared analysis unless the instrument is calibrated specially for the milk.