Abstract:
Animal Production in Australia PRELIMINARY REPORT ON THE ESTIMATION OF AVERAGE FIBRE DIAMETER OF GREASY WOOL BY NEAR INFRARED REFLECTANCE SPECTROSCOPY R.F. SCOTT*, E.M. ROBERTS*, SUMMARY A study is being made on the use of Near Infra Red Reflectance Spectroscopy (NIRS) in a fleece measurement system for sheep breeders. Comparisons between the standard airflow technique and NIRS for the estimation of fibre diameter in wool are reported. The airflow technique was compared with NIRS using four different methods of sample preparation. Correlations between airflow and NIRS estimates on greasy samples, minicored samples, scoured samples and scoured carded samples were 0-79., 0.59, 0.87 and 0.91. The method is rapid, allows simultaneous estimation of wool yield, and can be computer linked for ease of data handling. INTRODUCTION Two characters of importance in selecting for improved wool production are clean wool weight per head and average fibre diameter (Turner 1979). Fleece testing services provided by various Departments of Agriculture and The University of New South Wales have been able to provide objective estimates of these characteristics for sheep breeders for a number of years. The measurement systems used have been largely based on aqueous scouring for yield estimates and the airflow method for fibre diameter. The sample preparation involved for these methods is labour intensive. Connell and Norris (1981) consider a scoured yield estimate requires 1.5 worker hours. Experience at The University of New South Wales indicates that to produce an estimate of both yield and diameter requires 44 worker minutes per sample on average. The aqueous scour method requires a large wool sample (60g) and repeat yield measurements cannot be made on the same sample. Several workers (Scott and Roberts 1978; Connell and Norris 1981) have shown that NIRS can produce estimates of yield that are closely correlated with aqueous scoured yield. Furthermore, the NIRS process is rapid, requires minimal sample preparation, and is non-destructive, in the sense that repeat readings of the same sample can be made. A single yield estimate using NIRS requires between five and ten operator minutes per sample. Higgerson et al. (1981) looked at the estimation of wool base using NIRS, mwith the aim of producing an alternative to the aqueous scour method presently used to obtain a pre-sale test certificate. The results, although promising, have not achieved the accuracy required by IWTO test method specifications (IWTO 1976) for a large number of samples. The results obtained by Scott and Roberts (1978) were considered sufficiently accurate for use in a sheep-breeding guidance system, where the degree of accuracy required is not as high. The work reported here is a continuation of that work and looks at the use of NIRS in predicting average fibre diameter of greasy wool samples. Many workers in NIRS technology have shown the close correlation between NIRS estimates and measured particle size in powders (Johnson 1952; Melamed 1963; Schatz 1967). If a wool fibre 'is considered as a particle, the two variables affecting particle size will be fibre diameter and fibre length. Normally, fibre length is of a much greater magnitude than diameter thus changes in measured particle size will be closely related to fibre diameter. * School of'Woo1 and Pastoral Sciences, The University of New South Wales, P-0. Box 1, Kensington, N.S.W. 2033 M.J. KEOGH* 515 Animal Production in Australia MATERIALS AND METHODS Experiment 1 Onehundred samples of approximately 400 g each were randomly selected from a sale lot originating in the South East region of N.S.W. The samples were virtually free of vegetable matter. Although the aim of the work was to reduce the time involved in sample preparation, it was decided to use four different preparations in the hope that variations in the results would indicate possible sources of error in the NIRS estimates. The four methods of sample preparation were as follows; (a) greasy staple (no preparation); (b) Minicore*; (c) scoured staple; (d) scoured, carded staple. All sub-samples were randomly plucked from the original sample on a grid system to ensure a representative distribution of fibre diameter within the sub-sample. For the purposes of this work, triplicate airflow readings were used as absolutes against which to compare NIRS estimates, the correlation between triplicate airflow readings being 0.939. For each of the four preparations thirty samples were used in duplicate to calibrate the spectrophotometer. These calibration constants were used to predict diameter for a further 50 'unknown' samples. The use of 30 samples for calibration was recommended by the manufacturer and further work is being carried out to determine the validity of this. Each sample was read six times, the position of the sample being altered after each reading. For each sample (in duplicate), the NIRS fibre diameter (calculated from the average of the six readings) was compared with the average of the triplicate airflow diametersExperiment 2 An alternative to airflow measurement of diameter is available from the New South Wales Department of Agriculture's Trangie Research Station where flock rams are divided into diameter grades by means of liquid scintillation measurements (Downes 1971). One 10 g sub-sample was plucked on a grid system from each of the 100 samples. These samples were sent to Trangie where they were further sub-sampled to produce duplicate scintillation estimates of diameter for each sample. The average of the duplicate scintillation values was compared with the triplicate airflow estimates and asingleNIRS estimate. RESULTS Experiment 1 TABLE 1 Simple correlation coefficients for a single NIRS diameter estimate (produced from the average of 6 readings) on an unknown sample, compared with the mean of triplicate airflow diameter estimates Animal Production in Australia The correlations of the calibration samples are presented to verify that no bias was introduced by the method of their selection. The greasy staple (unknowns) and greasy staple repeat did not differ significantly, therefore, their correlations were pooled. Similarly, there was no significant difference between any of the duplicate results, therefore, all duplicate results were pooled. The pooled correlations for greasy, minicored, scoured and scoured/carded samples were 0.741, 0.522, 0.870 and 0.906 respectively. The correlations between the mean of the duplicate NIRS diameters and the mean airflow diameter for greasy sample and minicored samples were 0.788 and 0.906 respectively. A slight, though non-significant improvement in correlations occurred as a result of using the mean of two estimates, rather than a single NIRS diameter estimate. The minicore correlations were significantly lower than the greasy correlations. There was no significant difference between the correlations for scoured and scoured/carded samples, although carding always resulted in an increase in correlation. There was a significant difference between the scoured and the greasy preparations. Experiment 2 The correlations between the mean of the duplicate liquid scintillation results and the average of triplicate airflow diameters were calculated and compared with the correlations produced by the average of two NIRS estimates. TABLE 2 Simple correlations between the mean of duplicate NIRS diameters and the mean of triplicate airflow diameters The correlations from the greasy sample and greasy minicore preparations were significantly different (1%). The greasy sample and scoured preparations differed significantly at the 5% level. The results indicate the NIRS technique using either scoured preparation is as accurate as the scintillation technique in estimating fibre diameter. As a further means of comparison, the results for each sample were sorted into the fibre diameter grades used by the New South Wales Department of Agriculture's Fleece Testing Service (McGuirk 1978). The system is based on sorting animals into seven diameter grades, ranging from 3 pm finer to 3 urn stronger than the flock average in steps of 1 urn. Each sample was graded on its respective diameter as estimated by; (a) a single scintillation estimate, and (b) a single NIRS estimate on a greasy sample. TABLE 3 Comparison of the percentage of sheep which changed from the grades established by the mean of triplicate airflow diameters The results indicate that determination of fibre diameter grades by NIRS on a single greasy sample is as accurate as grades established by a single scintillation estimate. 517 Animal Production in Australia DISCUSSION Melamb (1963) in examining NIRS reflectance values for particles large in diameter in relation to wavelength, as is the case with wool fibres, was able to show a close relationship between reflectance values and particle size. The equation he produced relates the reflectance coefficient to the refractive index, the optical absorption coefficient, and the average diameter of the particles. For scoured wools of varying diameter, or wools of a constant yield and varying diameter the refractive index and absorption coefficient remain constant, thus the reflectance coefficient varies with diameter. The relationship appears to come about as a result of changes in the average angle of incidence of the beam on the fibre surface that occur as diameter varies, producing variations in the degree of scattering that results from the sample surface. The low correlations obtained for minicored wool may be a result of the reduction in fibre length that occurs with minicoring. Mean minicored fibre length (1.8 mm) is of a similar magnitude to fibre diameter, and predicted particle size would be related to both length and diameter, rather than diameter alone, as is the case with a full length fibre. Coatings of grease and other deposits on the fibre surface would also produce variations in the apparent diameter. A fibre from a low-yielding sample would appear to have a coarser diameter than actually occurs. This possibly accounts for the higher correlations that occur in both scoured and scoured/carded samples when compared with greasy samples. Further increases in accuracy would be expected in the case of carded wool as the carding process tends to remove extraneous vegetable and mineral matter. The surface thus presented to the instrument would consist almost purely of fibres oriented in a uniform direction. The results presented show the scoured/carded sample correlations to be consistently better thanthoseof the scoured samples though not significantly so. NIRS diameter estimates on greasy samples have been shown to be as accurate as scintillation estimates when used in a diameter grading system. When this is coupled with the fact that the instrument simultaneously produces a useful estimate of yield (Scott and Roberts 1978) and thus clean fleece weight, with greatly reduced costs and time per sample and the ability to feed results directly into a computer, then the potential of NIRS in a guidance testing system for sheep breeding is obvious, although the work presented here is limited and applies only to one flock. A large number of variables still remain to be examined. REFERENCES CONNELL, J.P and BROWN, G.H. (1978). J. Text. Inst. 12: 357. CONNELL, J.P and NORRIS, K.H. (1981). Text. Res. J. 51: 339. E DOWNES, A.M. (1971). Appl. Polymer. Symp. 18: 857. DOWNES, A.M. and TILL, A.R. (1968). Text. Es. J. 38: 523. HIGGERSON, G J ., MARLER, J.W. and CONNELL, J.P. IT& Technical Committee Report, Christchurch, April 1981. INTERNATIONAL WOOL TEXTILE ORGANISATION (1976). IWTO - 19 - 76 (E). JOHNSON, P.D. (1952). J. Opt. Soc. Am. 42: 978. McGUIRK, B.J. (1978). Wool Technol. Sheep Breed. 27: 17. MELAMED, N.T. (1963). J. Appl. Physics. 34: 560. = SCHATZ, E.A. (1967). In 'Modern Aspects of Reflectance Spectroscopy', p. 197 ed. W.W. Wendlandt. (Plenum Press: New York.) SCOTT, R.F. and ROBERTS, E.M. (1978). Wool Technol. Sheep Breed. 26: 27. TURNER, H.N. (1979). In 'Sheep Breeding', 2nd ed. p. 93, editors CL. Tomes, D.E. Robertson, R.J. Lightfoot. (Butterworths: Sydney.) 518