Revised Method for Aflatoxins in Cottonseed Products, and Comparison of Thin Layer and High Performance Liquid Chromatography Determinative Steps: Collaborative Study

Modifications to save analyst time and reagents have been made in the rapid method for separation and quantitation of aflatoxins in cottonseed products. A collaborative study of the revised method, including optional high performance liquid chromatography (HPLC) in addition to thin layer chromatography for the determinative step, showed no change in the method precision or accuracy due to the modifications and no significant difference in the values obtained by either determinative procedure. The HPLC technique provided no significant difference in analysis repeatability, but did reduce the betweenlaboratory error component. The method has been adopted as official first action.

Liquid Chromatographic Determination of Aflatoxin M1 in Milk Powder Using Immunoaffinity Columns for Cleanup: Interlaboratory Study

A liquid chromatographic method for determining low aflatoxin M1 concentrations in milk was evaluated in an International Dairy Federation (IDF) interlaboratory study. The study involved 16 participants from 11 countries. The method, chosen after a comparison of several methods by a preparatory group, uses an immunoaffinity column for cleanup. As the sample passes through the column, antibodies selectively bind with aflatoxin M1 (antigen) present and form an antibody-antigen complex. All other components of the sample matrix are washed off the column with water. Then, aflatoxin M1 is eluted from the column with acetonitrile, which is collected. Final determination is carried out by reversed-phase liquid chromatography with fluorescence detection. Over the tested range (80-600 ng aflatoxin M1/kg milk powder), an RSDR ranging from 11 to 23% was obtained by analyzing 24 samples (blind duplicates), 2 samples of which were blanks.

An analytical survey of aflatoxins in tissues from swine grown in regions reporting 1988 aflatoxin-contaminated corn

A joint project was undertaken by the Food Safety and Inspection Service (FSIS) and the Agriculture Research Service branches of the U.S. Department of Agriculture to determine the presence of aflatoxins in the U.S. meat supply during a drought year. In 1988, high incidences of aflatoxins occurred in corn grown in regions of the Midwest, Southeast, and South. Six states were identified as having serious aflatoxin contamination in their corn crop: Virginia, North and South Carolina, Texas, Iowa, and Illinois. Swine liver and pillars of diaphragm (muscle) tissues were sampled by federal FSIS Inspectors in plants located in these states. A worstcase sampling plan was conducted. Samples were taken in January 1989 from hogs fed corn soon after harvest and in April 1989 from hogs fed corn originally stored and then fed in the spring. A modification of the official AOAC method for the thin-layer chromatography (TLC) determination of aflatoxins in animal tissue was used to permit quantitation by LC with fluorescence detection. The official AOAC TLC confirmation of identity method was used to confirm all positive samples with B1 concentrations greater than 0.04 ppb and M1 concentrations greater than 0.1 ppb. Sixty samples in the January group and 100 samples in the April group were assayed. Concentrations of aflatoxins B1 and M1 in the first group of pig livers ranged from 0.04 to 0.06 ppb. The identity of aflatoxin B1 was confirmed in all positive samples. Aflatoxin M1 could not be confirmed in any of the positive liver samples because the method was insufficiently sensitive for this aflatoxin.(ABSTRACT TRUNCATED AT 250 WORDS)

Rapid immunochemical screening method for aflatoxin B1 in human and animal urine

A method has been developed to determine the presence of aflatoxin B1 in the urine of animals (including humans) by utilizing commercial immunochemical kits that can be used in the field. Urine is treated with diatomaceous earth and filtered to clarify the sample; 2-3 ppb aflatoxin B1, corresponding to about 300 ppb in the ingested feed/food, can be detected in the filtered urine without further purification. To improve sensitivity, the urine filtrate is passed through a C18 solid phase column to extract the aflatoxin. The column is washed with acetonitrile-water (15 + 85) and water, aflatoxin B1 is eluted with methanol-water (7 + 3), and water is added to the eluate, which is then tested for aflatoxin with the test kit. The limit of detection is 0.2 ppb, reflecting consumption of 40 ppb or more aflatoxin in the feed/food. When the initial sample volume is adequate, purification through the C18 column step is usually sufficient. For limited sample volumes, the eluate from the C18 column is mixed with water, added to an immunosorbent affinity column, and washed with water to remove excess sample matrix and impurities. Aflatoxin B1 is eluted with acetonitrile. The extract is evaporated under nitrogen and the residue is redissolved in methanol-water (25 + 75). At this purification stage, the limit of detection is reduced to 0.05 ppb.

Rapid Immunochemical Screening Method for Afiatoxin B1 in Human and Animal Urine

A method has been developed to determine the presence of afiatoxin B1 in the urine of animals (including humans) by utilizing commercial immunochemical kits that can be used in the field. Urine is treated with diatomaceous earth and filtered to clarify the sample; 2-3 ppb afiatoxin B1, corresponding to about 300 ppb in the ingested feed/food, can be detected in the filtered urine without futher purification. To improve sensitivity, the urine filtrate is passed through a C18 solid phase column to extract the afiatoxin. The column is washed with acetonitrHe-water (15 + 85) and water, afiatoxin B1 is eiuted with methanol-water (7 + 3), and water is added to the eluate, which is then tested for afiatoxin with the test kit. The limit of detection is 0.2 ppb, reflecting consumption of 40 ppb or more afiatoxin In the feed/food. When the initial sample volume is adequate, purification through the C18 column step Is usually sufficient. For limited sample volumes, the eluate from the C18 column is mixed with water, added to an immunosorbent affinity column, and washed with water to remove excess sample matrix and impurities. Afiatoxin B1 is eiuted with acetonltrile. The extract is evaporated under nitrogen and the residue is redissolved in methanolwater (25 + 75). At this purification stage, the limit of detection is reduced to 0.05 ppb.

Liquid chromatographic method for determination of citreoviridin in corn and rice

Citreoviridin, a neurotoxic mycotoxin, has been found as a natural contaminant in corn left unharvested in the southeastern United States and in rice of several Asian countries, including Japan. A reliable analytical method for the quantitative determination of citreoviridin in corn and rice is described. Corn or rice is extracted with dichloromethane, and the extract is partially purified on silica and amino solid-phase extraction (SPE) columns. The extract is analyzed for citreoviridin by normal-phase liquid chromatography, using a mobile phase of ethyl acetate-hexane (75 + 25) at 1.5 mL/min and a fluorescence detector to measure the yellow fluorescence (388 nm excitation, 480 nm emission). With a 100 microL injection loop, the relationship between concentration and injection volume is linear for 20-60 microL injections. Recoveries of citreoviridin added to yellow corn at 10-50 ng/g were 91.0-96.9%; recoveries from white corn (10-50 ng/g added) were 96.8-102.8%. Recoveries of 5000 ng/g added to white corn were 89.0%, indicating that heavily contaminated samples can be assayed by the method. Minimum detection limits were 10 ng for citreoviridin standard and 2 ng/g for citreoviridin added to corn. White rice fermented with Penicillium citreo-viride (1524 ppm) was mixed with and serially diluted with uncontaminated ground corn to obtain citreoviridin-contaminated corn (ca 25 ppb). When the samples were assayed by the method, a mean level of 24.4 +/- 1.65 ppb (6.5% coefficient of variation) was obtained. Four fermented rice food samples and 3 commercial rice samples were investigated.(ABSTRACT TRUNCATED AT 250 WORDS)

Citreoviridin levels in Eupenicillium ochrosalmoneum-infested maize kernels at harvest

Citreoviridin contents were measured in eight bulk samples of maize kernels collected from eight fields immediately following harvest in southern Georgia. Citreoviridin contamination in six of the bulk samples ranged from 19 to 2,790 micrograms/kg. In hand-picked samples the toxin was concentrated in a few kernels (pick-outs), the contents of which were stained a bright lemon yellow (range, 53,800 to 759,900 micrograms/kg). The citreoviridin-producing fungus Eupenicillium ochrosalmoneum Scott & Stolk was isolated from each of these pick-out kernels. Citreoviridin was not detected in bulk samples from two of the fields. Aflatoxins were also present in all of the bulk samples (total aflatoxin B1 and B2; range, 7 to 360 micrograms/kg), including those not containing citreoviridin. In Biotron-grown maize ears that were inoculated with E. ochrosalmoneum through a wound made with a toothpick, citreoviridin was concentrated primarily in the wounded and fungus-rotted kernels (range, 142,000 to 2,780,000 micrograms/kg). Samples of uninjured kernels immediately adjacent to the wounded kernel (first circle) had less than 4,000 micrograms of citreoviridin per kg, while the mean concentration of toxin in kernel samples representing the next row removed (second circle) and all remaining kernels from the ear was less than 45 micrograms/kg. Animal toxicosis has not been linked to citreoviridin-contaminated maize.

Optimum conditions for formation of aflatoxin M1-trifluoroacetic acid derivative

Because thin-layer chromatographic (TLC) confirmation of identity and reverse-phase liquid chromatographic (LC) determination with fluorescence detection of aflatoxin M1 both require the derivative formed in the reaction of M1 and trifluoroacetic acid (TFA), various reaction conditions were studied to obtain complete derivative formation. Of the various organic solvents tested, the reaction between M1 and TFA proceeded best in the nonpolar solvents hexane and isooctane. Other parameters investigated were reaction temperature and time, aflatoxin M1 concentration, and solvent volume. The following procedure is considered optimum: 200 microL each of hexane and trifluoroacetic acid are mixed with M1 standard in a silylated glass vial or with milk residue in a regular glass vial with a Teflon-lined screw cap and heated 10 min at 40 degrees C. The mixture is evaporated to dryness under N2, and the derivative is saved for TLC or LC. No unreacted aflatoxin M1 was detected by reverse-phase LC after this procedure was incorporated for analysis of milk samples.

Distribution and clearance of aflatoxins B1 and M1 in turkeys fed diets containing 50 or 150 ppb aflatoxin from naturally contaminated corn

Turkeys were fed a diet containing 50 or 150 ppb aflatoxin for 11 or 13 weeks or fed these diets for 11 weeks and then the control diet for 1 or 2 weeks. Aflatoxins B1 and M1 were found in liver, kidney, gizzard, and feces of poults fed the diets for 11 or 13 weeks. However, in turkeys fed the control diet for 1 or 2 weeks after the 11-week feeding trial, no residues of aflatoxin were found in the feces or tissues, except for some aflatoxin B1 remaining in detectable amounts in the gizzard. No mortality was attributable to aflatoxin, and there were no notable differences among groups in weight gains, feed conversion, or histopathologic changes in selected tissues. The response to a second inoculation with sheep erythrocytes was significantly lower in poults given dietary aflatoxin than in controls. This reduced antibody response was not observed when a Pasteurella multocida vaccine was administered.

Rapid liquid chromatographic determination of aflatoxins M1 and M2 in artificially contaminated fluid milks: collaborative study

An international collaborative study involving 14 collaborators from 5 different countries was conducted to test a rapid liquid chromatographic (LC) method for detecting aflatoxins M1 and M2 in fluid milk. Each collaborator prepared artificially contaminated milk samples (0.078-1.31 ng M1/mL and 0.030-0.13 ng M2/mL) by adding solutions containing various concentrations of aflatoxins M1 and M2 to fresh milk. Recoveries ranged from 85.2 to 102.5% (av. 93.7%) for aflatoxin M1 and from 99.5 to 126.7% (av. 109.8%) for aflatoxin M2. Coefficients of variation averaged 21.4% (M1) and 35.9% (M2). An analysis of variance was calculated from combined data to determine variance components. The within-laboratory variations (So) (repeatability) were 27.9% (M1) and 23.9% (M2), and the among-laboratory variations (Sx) (reproducibility) were 44.5% (M1) and 64.7% (M2). No visual differences were determined between normal or reverse phase LC for contaminated samples; however, there were an insufficient number of collaborators using normal phase to give meaningful separate statistical data. For 26 observations of uncontaminated milk, 3 false M1 positives were reported for normal phase LC determinations and 2 false M1 positives were reported for reverse phase LC determinations. Three normal phase and 11 reverse phase false M2 positives were reported for 104 observations in uncontaminated milk. The reverse phase LC method for determination of aflatoxins M1 and M2 in fluid milk has been adopted official first action.

Deoxynivalenol in hard red winter wheat: relationship between toxin levels and factors that could be used in grading

A study was made of deoxynivalenol (DON) incidences and levels in 1982 hard red winter (HRW) wheat grown in areas of Nebraska and Kansas known to have scabby wheat. Samples of wheat harvested in the areas were collected from elevators and analyzed for DON by gas chromatography with an electron capture detector. Of the 161 samples analyzed, 42% contained less than or equal to 1 ppm; 68% contained less than or equal to 2 ppm; 90% contained less than or equal to 4 ppm. There were differences in the occurrence of DON in the 5 areas identified in eastern Nebraska and Kansas. The mean level of DON decreased from north to south in these areas in the following order: 2.81, 2.73, 2.05, 1.52, and 0.83 ppm. An area in north central Kansas had a mean level of DON of 0.50 ppm. Correlations were made between DON incidences and levels in HRW wheat and factors used in grading wheat. The occurrence of DON was highly correlated with percent total kernels damaged by mold, percent total defects, and percent total scab damage.

Identification of aflatoxins by quadrupole mass spectrometry/mass spectrometry

MS/MS daughter experiments were recorded for aflatoxins B1, B2, G1, G2, M1, M2, and aflatoxicol, using 3 ionization modes. Daughters were recorded from the molecular ion (M+) using electron impact ionization (EI). Daughters from the protonated molecules (MH+) were recorded in the positive ion mode and the daughters from the molecular anion (M-) were recorded in the negative ion mode using chemical ionization (CI). These daughter spectra are all relatively simple. The EI daughters are quite similar to conventional EI spectra. The yield of (M-) is about 100 times greater than the yield of M+ in EI or MH+ in isobutane CI spectrum. Negative ion daughter spectra were used to demonstrate the feasibility of determining the presence of aflatoxin B1 in crude extracts of contaminated corn. Aflatoxin B1 could be detected at 10 ppb.

Gas chromatographic determination of deoxynivalenol in wheat

Modifications to a published method are described for the determination of deoxynivalenol (DON) in wheat by gas chromatography with electron capture quantitation of the heptafluorobutyrate derivative. In the modified method, DON is extracted by shaking the sample with methanol-water on a wrist-action shaker, followed by filtration through rapid flow paper. One concentration step is eliminated, and a hexane wash is incorporated to remove toluene from the silica gel column. Recoveries of DON from wheat samples spiked at 0.1, 0.5, and 1.0 ppm ranged from 77.3 to 86.3% and averaged 81.5%.

Fate of aflatoxins in tissues, fluids, and excrements from cows dosed orally with aflatoxin B1

A study was conducted to determine aflatoxins in tissues and non-tissues of 2 Holstein cows given oral doses of 0.35 mg of purified aflatoxin B1/kg of body weight/day for 3 consecutive days. Cow 1 was slaughtered 24 hours after the 3rd dose, and cow 2, after day 3, was fed aflatoxin-free rations for 7 additional days before slaughter. Tissue samples of brain, gallbladder and bile, heart, intestine, kidney, liver, lung, mammary gland, skeletal muscle, spleen, supramammary lymph nodes, thymus, and tongue, and nontissue samples of blood, feces, milk, rumen content, and urine were examined. Aflatoxins B1 and M1 were found in all samples of cow 1, except the thymus. Kidney, liver, and mammary gland had the highest concentrations of total aflatoxins (57.9, 13.2, and 25.1 ng/g, respectively), with the aflatoxin M1 concentration 40 times more than the aflatoxin B1 level in kidney. Aflatoxin residues were present (0.02 to 0.11 ng/g) only in kidney, liver, and intestine of the tissues from cow 2 (fed aflatoxin-free feed for 7 additional days). Aflatoxin B1 was not present in nontissue samples, but aflatoxin M1 (0.10 and 1.5 ng/ml) was found in the last milk and urine samples from the same cow. Urine assays are a possible way to monitor the presence of aflatoxin residues in meat tissues.

Effect of feeding corn naturally contaminated with aflatoxin on feed efficiency, on physiologic, immunologic, and pathologic changes, and on tissue residues in steers

Two of 3 groups of Holstein-Friesian steers (groups II and III; n = 5 each) were fed a ration containing corn naturally contaminated with 800 ng of aflatoxin/g. The other group of steers (group I; n = 5) was fed a ration containing noncontaminated corn. The respective rations were fed for 17.5 weeks, except the ration given to group III; the latter's first diet (contaminated with aflatoxin) was changed to a noncontaminated diet after 15 weeks, continuing for the remaining 2.5 weeks. All steers were killed and tissues and fluids were obtained for aflatoxin analysis. Although aflatoxin B1 and M1 could be detected in blood and urine at several sampling times during the experimental period in groups II and III steers (given the diets containing aflatoxin), there appeared to be no effects on body weight gains and immune phenomena, such as lymphoblastogenesis and antibody production, but there was a waning of the delayed cutaneous hypersensitivity in steers given aflatoxin-contaminated diets. In group III animals (diet was changed to noncontaminated ration at 15 weeks), aflatoxin B1 and M1 disappeared from urine before they were slaughtered. All tissues and fluids, except the rumen contents from these group III steers, were void of detectable aflatoxins B1 and M1 at necropsy. The concentrations of aflatoxin B1 in the rumen content of the latter steers were low. All tissues collected at necropsy from the group II steers fed the aflatoxin diet throughout the 17.5 weeks had detectable aflatoxins B1 or M1 present.

Determination and thin layer chromatographic confirmation of identity of aflatoxins B1 and M1 in artificially contaminated beef livers: collaborative study

An international collaborative study involving 13 laboratories was conducted to test methods for the determination and thin layer chromatographic (TLC) confirmation of identity of aflatoxins B1 and M1 in beef liver. For the determination, each collaborator furnished fresh or frozen beef liver. Samples were artificially contaminated by adding solutions containing various concentrations of aflatoxins B1 and M1 (0.032-0.69 ng/g). Two TLC confirmation methods were tested with extracts obtained from the determination. Two measurement methods using 2-dimensional TLC were evaluated. In the first, sample extracts were compared directly with B1 and M1 standards on TLC plates; in the second, internal standards plus sample extracts were compared with B1 and M1 standards on the plates. Average within-laboratory coefficients of variation (CV) for the direct method were 26% for B1 and 26% for M1 compared with 24 and 26%, respectively, for the internal standard method. The average between-laboratory CV values were 39% for B1 and 41% for M1 by the direct method and 36% for B1 and 39% for M1 by the internal method and 36% for B1 and 39% for M1 by the internal standard method. Recoveries ranged from 64 to 90% for B1 and from 72 to 86% for M1. These data indicate that the more convenient direct method was sufficient, and internal standards were unnecessary. An analysis of variance was calculated from combined sample data to determine components of variance. The within-laboratory CV values were 27.0 and 32.3% for B1 and M1, respectively, and the between-laboratory CV values were 47.1 and 53.2%, respectively. Both TLC confirmation methods gave satisfactory results and have been adopted official first action, along with the determination method.

Determination of aflatoxins in animal tissues

A method for the determination of aflatoxins in animal tissues has been developed, and applied successfully to beef, swine, chicken, and human livers, and to beef kidney, heart, spleen, muscle, and blood. Blended tissue is denatured with citric acid and extracted with dichloromethane on a wrist-action shaker. After filtration, the extract is partially purified on a silica gel column, and aflatoxins B1 and M1 are determined by 2-dimensional thin layer chromatography and densitometry. Recoveries of B1 and M1 added to meat tissues and blood were approximately 90 and 80%, respectively. The method gave results for a contaminated freeze-dried liver comparable to analyses by 3 other published meat tissue methods. The method is rapid and has a determination limit less than or equal to 0.1 ng/g. In addition, the method uses less toxic and smaller quantities of solvents and chemicals.

Improved method for confirmation of identity of aflatoxins B1 and M1 in dairy products and animal tissue extracts

A method is described for confirming the identity of aflatoxins B1 and M1 in dairy products and liver extracts on a thin layer plate. Extracts and standards containing aflatoxins B1 and M1 are spotted on 10 x 10 cm plates, which are developed 2-dimensionally in mixtures of isopropanol-acetone-chloroform. After the first development, trifluoroacetic acid-hexane (1 + 4) is sprayed on that part of the plate containing the separated extract components and the underdeveloped standard spots of B1 and M1, and the plate is heated 6-8 min at 75 degrees C. Then the plate is developed in a second direction, and the reaction products of B1 and M1 with trifluoroacetic acid from the extract are compared with the same derivatives of the respective standards. The method has been used successfully on extracts of milk, cheese, and liver containing 0.1 ng B1 or M1/g and can be completed in 35-45 min.

Aflatoxins M1 and M2 and Parasiticol: Thin Layer Chromatography and Physical and Chemical Properties

A survey of known anatoxin solvent systems for resolution of aflatoxins M1 and M2 on TLC plates revealed that the best system for determining aflatoxins B1, B2, G1, G2, M1, and M2 is isopropyl alcohol-acetone-chloroform (5 + 10 + 85). Substitution of various alcohols for isopropyl alcohol in this system demonstrated that maximum resolution of M1 and M2 was achieved with n-amyl alcohol-acetone-chloroform (10 + 10 + 80); however, B1, B2, G1, and G2 migrated with the solvent front. When alcohol-chloroform (5 + 95) mixtures were investigated, n-propyl, n-butyl, and tert-butyl alcohol + chloroform resolved M1 and M2 best but did not separate B1, B2, G1, or G2. Molar absorptivities of both M1 and M2 were determined in methanol, chloroform, acetonitrile, and acetonitrile-benzene (2 + 98). Relative fluorescent intensities of aflatoxins B1, M1, and M2 were compared on both developed and undeveloped TLC plates. Fluorescent intensities of B1 and M1 on silica gel were nearly equal, and the intensity of M2 was 1.4–1.5 times that of the other 2 aflatoxins. Water adducts of anatoxin M1 and parasiticol were prepared. The diacetate adducts of parasiticol were formed by treatment with acetic anhydride and concentrated HC1. Monoacetyl derivatives of M1, M2, and parasiticol were obtained by treatment with pyridine and acetic anhydride. Good resolution of the water-addition derivatives of B1, G1, parasiticol, and Mi on TLC plates was achieved with isopropyl alcoholacetone-chloroform (5 + 10 + 85).

Collaborative Study of the Determination of Aflatoxin in Corn and Soybeans

A study was made to determine whether the procedure commonly designated as the CB method, previously recommended for determining anatoxin in peanuts and peanut products, could be applied to corn and soybeans. Both naturally contaminated and spiked corn and soybeans were extracted and analyzed by TLC. Of the 15 collaborating laboratories, 9 were equipped to measure anatoxins on TLC plates densitometrically, as well as visually. Seven collaborators used molar absorptivities to determine the concentration of anatoxins in solutions to be used for TLC standards. The official CB method has been extended to include corn and soybeans, and the rapid TLC screening technique for aflatoxin in corn has been adopted as official first action.

Aflatoxins: Improved Resolution by Thin Layer Chromatography

Water was added to solvent systems for TLC of aflatoxins to achieve more reproducible results in laboratories where temperature and relative humidity vary. Resolution of the toxins also improved. Increments of water were added to solvent systems composed of acetone-chloroform (10 + 90, 12 + 88, and 15 + 85, v/v). As the water concentration was increased, separation of aflatoxins B2 and G1 improved. These two toxins are usually the most difficult to resolve in these solvent systems. Separations were the best with wateracetone-chloroform (1.5 + 12 + 88, v/v/v). Water added to methanol-chloroform (3 + 97, v/v) improved resolution of this solvent system but not enough for quantitative analysis, or at times, qualitative analysis. The solvent system water-methanol-ether (1 + 3 + 96, v/v/v) separated aflatoxins as well as water-acetone-chloroform (1.5 + 12 + 88, v/v/v).

Production of aflatoxin on wheat and oats: measurement with a recording densitometer

A method has been developed for the production of aflatoxin by growing Aspergillus flavus NRRL 3145 on solid substrate wheat. Optimal yields of 900 mug of aflatoxin G(1) and 900 mug of aflatoxin B(1) per g of substrate were obtained in 4 to 5 days at 28 C. A study of aflatoxin production on hulls and groats of oats and on whole oats by A. flavus strains NRRL 2999, NRRL 3000, and NRRL 3145 revealed that aflatoxin was produced on all three substrates, although production was very slight on hulls. Strain NRRL 3145 grown on solid substrate groats produced the largest amounts of aflatoxin: 580 mug of B(1) and 450 mug of G(1) per g of substrate. A densitometric method for reading thin-layer chromatographic plates is described; this is more objective and more accurate than the visual methods previously used for the determination of all four aflatoxins.

Production of aflatoxin on rice

A method has been developed for the production of aflatoxin by growing Aspergillus flavus strain NRRL 2999 on the solid substrate rice. Optimal yields, more than 1 mg of aflatoxin B(1) per g of starting material, were obtained in 5 days at 28 C. A crude product containing aflatoxins was isolated by chloroform extraction and precipitation with hexane from concentrated solutions. The crude product consisted of 50% aflatoxin in the following ratio: B(1)-B(2)-G(1)-G(2), 100:0.15:0.22:0.02. Aflatoxin B(1) was separated from almost all the impurities and from the other aflatoxins by chromatography on silica gel with 1% ethyl alcohol in chloroform. Analytically pure aflatoxin B(1) was recrystallized from chloroform-hexane mixtures.