Some applications of near infra-red reflectance (NIR) analysis in plant breeding have been tested. Prediction of the nitrogen content of cereals, kale and beans, and nitrogen and oil in rapeseed was sufficiently accurate for screening purposes although wheat straw gave a less accurate result. The prediction of nitrogen content of wheat grain could be supplemented by the simultaneous estimation of grinding resistance and bran cleanliness, characters of importance for bread-making quality. Hagberg falling number, sodium dodecyl sulphate (SDS) sedimentation volume and loaf volume could not be accurately predicted by NIR. Encouraging results were obtained for predicting malting quality of barley and the digestibility of kale samples.

A dry-block digestion system suitable for the estimation of total protein nitrogen in plant material has been evaluated. The method was found to be as accurate as Kjeldahl flask digestion, required much less fume cupboard space and enabled samples to be dealt with in large batches through weighing, digestion and analysis. The batch size can be designed to be compatible with the autoanalyser capacity. These advantages result in more rapid analysis with less chance of errors and accidents. A 45 min digestion period was found suitable for several crop species of widely different protein content, only field bean seeds required longer digestion time. Using the technique described samples are digested at a relatively low temperature (330 °C) for a short time and digests do not solidify on cooling. The autoanalyser method for determining ammonia concentrations in barley and malt digests (Micheson & Stowell, 1970) has been slightly modified for use with the plant materials studied. The dry-block and autoanalyser system is rapid, precise and compact, requires small amounts of tissue and is suited to the routine analysis of breeding material.

A method is described in which glucosinolates are simultaneously extracted from crushed seeds and degraded by myrosinase. Glucose released by myrosinase is measured by an autoanalyser using a single reagent.Using conditions optimized for samples of 200 mg, yields were slightly greater than those obtained from corresponding samples of defatted meal. Coefficients of variation of about 4% were obtained over a range of more than 100/miol of glucosinolate/g seed. It is suggested that the method will be of value to plant breeders selecting from populations having low and intermediate concentrations of glucosinolate and to others wishing to obtain a rapid but precise measurement of these important compounds.

SummaryPredictions of nitrogen, oil and glucosinolate concentration in rapeseed samples were made by near infrared reflectance analysis after various grinding treatments. Also examined were the effects of normalizing reflectance data and the possible advantage of using all combinations of two and three wavelengths in the calibration regression analysis over forward stepwise regression. The main conclusion was that drying the samples prior to a controlled grinding treatment gave the best results, although acceptable results for selection purposes could be obtained using whole seeds to predict nitrogen and oil. None of the treatments of the seed or reflectance data allowed acceptable prediction of glucosinolate content.

CHEMISTRY extent, is of sufficient promise to justify further work with the method. Plans are being made to provide automatic temperature control, so that the test may be run overnight and hence reduce the time of test in terms of work days. ACKNOWLEDGMENTAcknowledgment is made to the laboratories which assisted in this investigation by supplying the oils tested. LITERATURE CITED(1) Coordinating Research Committee, "CRC Handbook," CRC