Chapter 6 Oil Content Analysis: Myths and Reality V.J. Barthet and J.K. Daun Canadian Grain Commission, Grain Research Laboratory, Winnipeg, Manitoba R3C 3G8, Canada Abstract The FOSFA (Federation of Oils, Seeds and Fats Associations Limited) extraction method (harmonized as AOCS Am 2-93 or ISO 659) is considered to be the refer- ence method to measure the oil content of oilseeds. The method is based on the extraction of neutral lipids with hexane or petroleum ether, and the extracted com- ponents are estimated gravimetrically and defined as crude fat or oil content. The method requires a triplicate grind/extraction, making it lengthy and very detailed for the analyst. Compared with other official methods (AOAC 996.06), the FOSFA extraction method gave the highest oil recoveries in all oilseeds tested. Accelerating the extraction through the use of new instrumentation, which com- bines solvent extraction with a physical disruption such as pressure or microwave heating, has not been able to give oil recovery equal to the FOSFA extraction. Although the FOSFA method remains the reference method, both ISO and AOCS have identified a need to develop new rapid methods for oil extraction that yield oil equivalent to that from the FOSFA method and that can be used for determina- tion of other factors such as methyl esters or free fatty acids. This study examined the lipid components extracted by each stage of the FOSFA method: First Extraction (FOSFA1), Second Extraction (FOSFA2), and Third Extraction (FOSFA3). In a typical analysis, FOSFA1 accounted for ~86%, FOSFA2 for ~13.5%, and FOSFA3 for as much as 0.5% of the total oil extracted. Although mainly triacylglycerols (TAG) were found, the extracts contained small amounts of other lipid components including nonesterified fatty acids, partial glycerides, fat-soluble vitamins, long-chain alcohols or aldehydes, and sterol or cholesterol esters. More nonpolar material was extracted in the first early stage of the extrac- tion, whereas the later stages contained more polar material. The oil in FOSFA1 and FOSFA2 contained close to 98% TAG based on methyl ester determination. The analyses of the phosphorus content by graphite furnace atomic absorption spectrophotometry showed that almost no phosphorus (below limit of detection) was found in FOSFA1 and FOSFA2 oils but small amounts of phosphorus were present in FOSFA3 oils. This indicates that only ~0.5% of phospholipids were pre- sent in the FOSFA3 oils, suggesting that only very small amounts (<0.005%) of phospholipids are extracted by the total method when the three extracts are com- Copyright © 2004 AOCS Press bined. Moreover, during the extractions of canola by the FOSFA method, the pro- portion of triacylglycerol (TAG) with n-7 fatty acids increased with each step of the process. These fatty acids are associated with the seed coat in canola seeds, and the increase suggests that these structural lipids are the last lipids to be extracted. Introduction The term oil refers to a mixture of lipids that is liquid at room temperature and usu- ally of plant origin. According to Gunstone and Herslöf (1), oil content is the ana- lytical quantity of oil obtained from plant sources (seed or endosperm). It is defined by both the source material and the extraction procedure. Different oil con- tents may be obtained from the same seed sample if the extractions use organic sol- vents of different polarity and/or different conditions of pressure and temperature. For example, if other conditions are the same, using ethyl ether as the extraction solvent will likely give a higher oil content than using petroleum ether because the more polar ethyl ether extracts more polar lipids. 1 For an oil processor, there is a direct relation between seed oil content and economic value. For soft oilseeds such as canola, the relative value of oil to meal may be as large as 8:1. If canola oil is sold at $500/tonne, the difference in value obtained between a tonne of seed at 40% oil and a tonne of seed at 41% oil is ~$5. For a nutritional scientist, oil content is related to energy intake and is part of the description of the nutritive value of the product. For a food scientist, oil content is linked to important functional and organoleptic qualities of the food. For a crusher, the oil content gives an estimate of the yield of a given batch; therefore, the oil content is an economic characteristic of the oilseed trade. To satisfy all par- ties involved in the oilseed trade, it is important to have an analytical method that gives the real oil content of the seeds. Moreover, it is important to have this (these) methods standardized. For some researchers, however, it might be acceptable to give up a little bit of accuracy and precision in favor of a very rapid method. For example, plant breeders continually seek rapid methods for estimating oil content. In the 1930s, researchers at the Canadian Grain Commission (2) developed a method using the refractive index of a hexane extract of flax seed. Others used the relationship between seed density and oil content (3) or the amount of oil expressed onto paper by a hydraulic press (4). For rapeseed, the first really useful rapid method for oil content determination was the method developed by Troëng (5), which was used at Svalöv to analyze as many as 60,000 samples/y (6). The method was adopted in Canada as the “Swedish” method. A study carried out by the Associate Committee on Grain Quality in 1962 and 1963 1 It is important that the reader be aware of the difference between petroleum ether (also known as com- mercial hexane, or by commercial names such as Skellysolve TM ) and diethl ether. Petroleum ether is a mixture of alkanes ranging from C 5 to C 8 and including branched chains defined by a boiling point range (usually between 40 and 60°C). Diethyl ether is a true ether and is considered more polar in nature than petroleum ether. Copyright © 2004 AOCS Press showed that the “Swedish” method recovered at least 1% more oil than the single grind method used at the Grain Research Laboratory (Table 6.1). These rapid extraction methods were replaced later by methods employing spectroscopic techniques such as nuclear magnetic resonance and near infrared spectroscopy. There is continued interest, however, in rapid extraction methods that will give an oil and meal that can be used for further experimentation. A rapid extractor, based on the Svalöv method, for small samples of seed was developed by Agriculture Canada in the 1980s (7). More recent techniques such as supercriti- cal fluid extraction (SFE) and accelerated solvent extraction have been proposed for the measure of oil content. It is necessary for results from these new methods to be compared with results from the reference method for oil content so that the pro- posed new methods may be validated.The FOSFA (Federation of Oils, Seeds and Fats Associations Limited) extraction method, harmonized as AOCS Am 2-93 and ISO 659, is considered the reference method with which to measure oil content in oilseeds by the international oilseed trade. This chapter examines the techniques and apparatus associated with the refer- ence method for oil content extraction. It also provides information on the compo- sition of the oil extracted by the reference method and thus provides a benchmark for studies using alternative methodologies. Some comparisons with other method- ologies are provided, particularly the AOAC method commonly used for food labeling and the AOCS SFE method. The relation between results from the refer- ence method and the efficiency of commercial extraction is also discussed. Soft and Hard Oilseeds. Oilseeds may be divided into groups that include soft oilseeds, such as sunflower, safflower, canola, rapeseed, flax, and sesame, and hard oilseeds, such as soybean and cottonseed. Certain oilseeds are more difficult to extract than others (8). Soybeans, for example, are an example of an oilseed from which it is relatively easy to extract the oil. Within the soft oilseeds, it is easier to TABLE 6.1 Comparison of Ball-Milling and Simple Extraction Procedures for the Determination of Oil Content in Rapeseed. Study by the Associate Committee on Grain Quality a 1962 1963 Method (%) Swedish b Mean 43.45 39.84 SD 0.74 0.30 Single grind/extraction c Mean 42.30 38.70 Difference 1.15 1.14 a Reported in minutes of the 47th and 48th meetings of the Canadian Associate Committee on Grain Quality, Appendices C14 and C11, respectively. b Six laboratories in 1962, five in 1963. c Results from the Grain Research Laboratory. Samples were ground on a flax mill (AOCS Af 3-54) and extracted for 8 h with petroleum ether on a Goldfisch extractor . Copyright © 2004 AOCS Press remove the oil from relatively large seeds such as sunflower and safflower than from smaller seeds such as rapeseed, sesame seeds, or flaxseeds. Particle size dis- tribution has an important effect on the efficiency of the oil extraction and its rate. Regrinding to reduce the particle size is a key process in the FOSFA extraction. Small soft oilseeds may also be difficult to extract because the particle size has to be smaller than their cell size so that the oil present in their cell structure is released. The authors studied SFE to measure oil content and observed an impor- tant matrix effect in soft oilseeds (9). Apparatus for Extraction Methods of Oil Content Analysis Several extractors are used to perform fat or oil extractions. The use of some extractors is recommended by some official methods, whereas others are strictly proscribed. The most common and unavoidable systems used for oil extraction in an analytical laboratory are described in this chapter. Butt Tube. In the Butt tube system, the sample is ground and a weighed portion is placed into a porous thimble or folded into filter paper. The thimble or filter paper is then placed into the Butt tube and the solvent is placed in the flask. The apparatus is assembled as shown in Figure 6.1 and the solvent is boiled; vapors rise to the con- denser where they condense and drip down through the sample back into the boiling solvent below. The extraction process is continuous and can be completed within a few hours, although exhaustive methods call for regrinding of the sample after the ini- Fig. 6.1. Left: Butt tube extractor. Right: Twisselman modification. Water-cooled condenser Tapered cork stopper Extraction tube Sample thimble sets here Tapered cork stopper 50- or 100-mL Soxhlet flask Copyright © 2004 AOCS Press tial oil has been removed and again near the end of the extraction. Once the extraction is finished, the solvent is removed from the extracted oil by distillation followed by vacuum drying. The extracted oil is weighed. The Goldfisch Fat Extractor (Labconco) has long been recognized as a commercial version of the Butt tube system. An improvement on the Butt tube is the Twisselman extractor, which adds a stopcock between the sample container and the condenser. When the stopcock is closed, it is possible to reclaim the solvent from the extracted material. The Twisselman extractor is recommended in the German standard method (DGF B-I.5) (88). Soxhlet Extractor. In a Soxhlet extractor (Fig. 6.2), the solvent is heated in a boil- er; the pure vapor rises up through a by-pass and into the top part of the Soxhlet container where the sample to extract is contained. In the condenser, the vapors are condensed and drip into the sample-containing thimble. When the level of liquid reaches the same level as the top of the siphon, the liquid containing the extracted material is siphoned back into the boiler. Soxhlet extraction is not a continuous procedure, but a batch system with repeated extractions. Usually a minimum of 30–50 cycles is considered necessary to complete the extraction. The use of a Soxhlet extractor is not recommended by the FOSFA method for oil content analysis. The temperature of the solvent in the solvent vessel rises dur- ing the extraction due to the presence of higher concentrations of oil in the solvent. Eventually, the pure solvent siphoning into the hot extract may vaporize very rapidly (“bump”), flooding the condenser and leading to a loss of oil and poor recoveries. Fig. 6.2. Soxhlet apparatus. Condenser By-pass tube Sample chamber Siphon Receiving flask Copyright © 2004 AOCS Press Immersion Extractors. Immersion extractors (Fig. 6.3) include the Soxtec TM sys- tem manufactured by Foss-Tecatur and, more recently, a system manufactured by Velp Scientificat. The original apparatus was designed by E. Randall and further developed by Foss-Tecator. The method is a two-step procedure based on the best parts of the Soxhlet and Twisselman methods. At the beginning of the extraction, the ground sample is placed in a thimble and lowered into the boiling solvent where the extractable material passes rapidly into the solvent. During the second phase of the extraction, the sample is raised from the solvent and the condensed vapors of the boiling solvent drip through the thimble containing the sample. In this rinse position of the apparatus, there is a continuous supply of fresh solvent, allowing the continuous increase of the concentration of extracted material in the solvent recovered in the cup. Once the extraction is completed, the valve on the drip tube may be closed so that the solvent can be recovered from the extract. Wet Ball Mills (Tröeng Extractor, Dangoumeau TM Tubes). Ball mills are also known as centrifugal or planetary mills (Fig. 6.4). For oilseeds, two ball mills are currently in use. The ball mill proposed by Tröeng (5), and known in Canada as “Swedish Tubes,” has been used in the ISO Method 10519 for determination of chlorophyll and in the AOCS Method Am 2-93 (1997) for determination of oil content (FOSFA) method. Unfortunately there do not appear to be any current commercial suppliers for either mill, although the Tröeng apparatus can be manu- factured by an efficient metal working shop. The use of a wet milling technique for grinding soft oilseeds reduces particle size to <100 µm, resulting in efficient extraction of the oil. In the original Tröeng method, which was designed primarily for plant breed- ing purposes, up to 3 g of whole seeds of small soft oilseeds were ground directly in an accurately dispensed volume of petroleum ether. After 45 min, the tubes were Fig. 6.3. Stages of operation of the immersion extractor. Immersion Rinsing Solvent recovery Copyright © 2004 AOCS Press set in an upright position and left for several hours until the fine flour had settled to the bottom. An accurate aliquot was then placed in a weighed beaker and the sol- vent removed by evaporation followed by vacuum drying. The beaker was weighed and the oil content was calculated gravimetrically, taking into account the contribution of the oil to the volume of the aliquot taken. A trained analyst could achieve good accuracy and precision using this method. The method was adapted to be used in extraction procedures as described above. Effect of Particle Size. Grinding the sample is one of the most important steps in oil content analysis. An excellent description of oilseed grinders is provided in the second chapter of the classic book on oil analysis by Mehlenbacher (10). It is nec- essary to reduce the particle size to as small as possible so as to disrupt the oil- bearing cells and to allow the extraction solvent the opportunity to percolate through the sample and contact all of the lipid material. An initial particle size of ≤2 mm for small seeds and 4 mm for large seeds has been recommended (8). For most efficient extraction, the particle size should be reduced during the procedure by microgrinding. A sufficiently small particle size is difficult to achieve with a single grinding step. Coupling a blade or impact mill as the first step, followed by wet ball milling with the Tröeng apparatus (Table 6.2) efficiently reduced the par- ticle size of canola seeds to the required level. After the third grind, >70% of the particles were <75 µm. In comparison, dry ball milling produced only 25%, grind- ing with diatomaceous earth only 33%, and two grindings with a blade-type coffee mill produced only 44% of particles with particle size <75 µm. The ISO and IASC methods require the use of 80- to 100-mesh silver sand as an adjunct to grinding, whereas the AOCS method for sunflower seed recommends grinding in the presence of diatomaceous earth. One of the objectives of using sand or diatomaceous earth is to scavenge oil that might be expressed in the grinding operation. The use of wet ball mills reduces the loss of oil on the grinding appara- tus because the apparatus is rinsed with the extraction solvent. Fig. 6.4. Ball mills used for wet grinding of oilseeds. Top, according to Tröeng (Tröeng 1955 #16020); bottom, Dangoumeau TM Ball Mill (ProLabo). Copyright © 2004 AOCS Press TABLE 6.2 Effect of the Grinding Procedure on the Particle Size of Canola (%) FOSFA or AOCS Am 2-93 (with ball mill) Coffee mill (first grind) Wet ball mill (second grind) Wet ball mill (third grind) Particle size (µm) B. rapa B. napus Canola B. rapa B. napus Canola B. rapa B. napus Canola >500 22.1 13.1 15.0 1.4 1.0 0.8 0.3 0.3 0.1 355–500 31.4 24.0 26.9 6.8 9.3 2.3 4.0 0.8 0.5 212–355 17.7 27.1 26.4 12.2 14.0 7.8 7.2 2.6 150–212 5.9 10.0 9.5 10.3 10.7 8.5 5.5 4.8 2.7 75–150 5.9 11.0 11.0 23.2 17.9 22.3 12.9 11.7 8.8 37–75 5.4 6.6 7.1 20.0 16.9 20.7 20.1 16.6 21.7 <37 12.5 8.3 4.2 26.0 30.3 37.7 50.0 63.4 63.5 Mikro-Samplmill Repeated Dry ball milling with diatomaceous earth coffee mill Particle size (µm) First grind Second grind Third grind (AOCS Ai 3-75) First Second >500 12.6 9.8 6.1 7.0 6.4 0.0 355–500 26.2 28.1 25.1 7.7 9.2 2.1 212–355 26.7 27.4 29.2 17.8 23.8 16.5 150–212 9.9 8.8 9.6 17.9 18.1 19.8 75–150 9.9 7.0 5.2 16.5 14.1 17.3 37–75 8.5 5.9 6.8 33.1 28.4 44.3 <37 6.2 12.9 18.0 Copyright © 2004 AOCS Press Oil Recovery and Precision of Single vs. Multiple Extractions. Despite evi- dence to show that multiple extractions are necessary to ensure that the oil recov- ery from oilseeds is complete, methods have been developed that require only a single grind and extraction step. An example of such a method was the AOCS Method Ai 3-75 developed originally for sunflower seed but adopted for canola and rapeseed in the early 1980s. This method required that the sample be ground in the presence of diatomaceous earth, which was thought to provide not only a supe- rior grind, but also to scavenge expressed oil from the grinding apparatus. As a part of the Smalley Check Sample program, AOCS asked laboratories to test canola samples by both the single grind method (Ai 3-75) and the (then) proposed FOSFA multiple grind method. Results (Table 6.3) showed that the FOSFA method not only gave a higher oil recovery than the single grind method but that the precision of the FOSFA method was superior to the precision of the single grind method. Official Methods FOSFA Extraction Method (AOCS Am 2-93 and ISO 659) The FOSFA extraction method is based on a three-stage grinding and extraction of neutral lipids with hexane or petroleum ether. The extracted component is then TABLE 6.3 Comparison of Oil Recovery and Precision for AOCS Ai 3-75 (1995) Single Grind with Diatomaceous Earth and FOSFA [AOCS Am 2-93 (1997)] Triple Grind Methods by Smalley Check Sample Participants (%) Comparison of recoveries AOCS FOSFA No. of Smalley sample no. mean mean Difference Paired tP-value samples 1 42.41 42.85 0.44 –1.350 0.202 13 2 42.59 43.13 0.47 –1.500 0.159 13 3 41.72 42.33 0.63 –2.360 0.033 15 4 42.66 42.88 0.22 –0.680 0.500 13 5 42.70 43.06 0.36 –3.480 0.005 13 All 42.40 42.82 0.42 –3.520 0.001 67 Comparison of precision No. of Smalley sample no. AOCS SD a FOSFA SD b Ratio F-value P-value samples 1 1.48 0.93 1.59 2.55 0.118 13 2 1.29 0.34 3.79 12.42 0.002 13 3 0.98 0.43 2.28 4.96 0.005 15 4 0.98 1.09 0.90 1.61 0.400 13 5 0.59 0.59 1.00 1.00 0.990 13 All 1.11 0.77 1.44 2.10 0.002 67 a A reproducibility SD of 0.78% was established for sunflower seed in AOCS Ai 3-74 (95). b A reproducibility standard deviation of 0.56% was established for rapeseed in AOCS Am 2-93 (1997). Copyright © 2004 AOCS Press estimated gravimetrically and defined as crude fat or oil content. The triplicate extraction with grind and regrind makes the method time consuming and requires great attention to detail on the part of the analyst. In comparison with other meth- ods, the FOSFA extraction method gives the highest oil recoveries in all oilseeds tested. Recoveries of oil at each step of the FOSFA method (Table 6.4) give an indi- cation of the rate of oil extraction and therefore an indication of the complexity or ease of this extraction for various oilseeds. The ease with which oil can be recov- ered from a particular oilseed can be estimated by the percentage of oil recovered in the first extract. The combined first two extracts for all oilseeds recovered 99% of the total oil. The third extract, except for yellow mustard, on average, recovered <1% of the total oil. The seeds from this study can be classed into three groups according to the level of oil recovered in the first extract: (i) recoveries >90% in the first extract (soybeans, the only hard oilseed in the study); (ii) recoveries ranging from 80 to 90% (solin, sunflower, flax, safflower, canola, and oriental mustard); all soft oil- seeds; (iii) recoveries <80% [brown mustard (Brassica juncea) and yellow mustard (Sinapis alba)], soft seeds that were also cited as difficult to extract in the SFE study (9). The hardness of the seed, soft oilseed vs. hard oilseed, plays an important role in the effectiveness of the first extract and therefore in the difficulty of the oil extraction. It was also suggested that the seed size plays a major role in oil extrac- tion recoveries (8), but the present results show that seed size had only a small effect on the oil recovery. Although there is a significant size difference between sunflower or safflower and canola or flax, the difference in the oil recoveries of the first extract was negligible. More importantly, there appears to be a matrix effect on the oil extraction recovery. This matrix effect might be due to variables such as hardness of the seed and oil body sizes or the oleosin and mucilage contents. Some extracts from the study were converted to fatty acid methyl esters (FAME) by base-catalyzed derivatization. In this transmethylation, the free fatty acids (FFA) are not converted into FAME. Although this method leads to an underestimation, it will be negligible in good quality oilseeds because the FFA are <1%. The method is considered the best and the fastest method with which to obtain FAME from oilseeds (11). Along with several other oilseeds, Brassica napus contains a 18:1(n-7) isomer along with the normal oleic acid 18:1(n-9) isomer. This fatty acid has been shown to be the major 18:1 isomer in the fatty acid composition of the seed hull (12). The separate hull and meat contents of the n-7 isomer were found to be independent of the extraction stage (Table 6.5) but when the seed was extracted, the n-7 isomer increased especially in the third extraction, suggesting that a significant portion of the structural lipid of the hull was extracted during this last stage. By including an internal standard (C 17 ) in the GC analysis, it was possible to demonstrate that the first two extracts contained between 100 and 93% acylglycerides Copyright © 2004 AOCS Press [...]... Safflower Soybean 114 43 .6 85.1 100 67 .8 7.8 14.2 30.8 0 7.5 99.3 0.8 1 .6 0 0.4 3 5 3 26 3 20 10 12 37.9 41.4 29.7 40.3 41.4 45.4 33.9 20.2 53 .6 81.7 76. 0 86. 7 88.5 87.9 85.7 94.8 73.1 83.1 76. 1 96. 2 89 .6 94.5 99.1 95.9 15.4 80.1 75.9 77.7 87.8 76. 9 59.5 93.8 33.1 1.3 0.2 7.9 0.9 6. 1 15 .6 0.7 45 .6 17 .6 23.0 13.0 10.9 11 .6 14.1 4.4 83.7 19.2 23.2 22.1 12.0 22.2 40.3 5 .6 26. 5 16. 3 22.8 3.4 9.2 4.5 0.9... TABLE 6. 5 Relative Composition in 18:1(n-7) and 18:1(n-9) Acids from Canola Hull, Meat, and Seed at Each Step of the FOSFA Extraction Relative composition total fatty acid profile 18:1(n-9) 18:1(n-7) Sample Hull Extraction 1 Extraction 2 Extraction 3 Meat Extraction 1 Extraction 2 Extraction 3 Seed Extraction 1 Extraction 2 Extraction 3 Relative composition in 18:1 18:1(n-9) 18:1(n-7) (%) 22.5 22.0 22 .6. .. Friedrich, J.P., List, G.R., and Heakin, A.J (1982) Petroleum-Free Extraction of Oil from Soybeans with Supercritical CO2, J Am Oil Chem Soc 59: 288–292 15 Eggers, R (1985) High Pressure Extraction of Oil Seed, J Am Oil Chem Soc 62 : 1222–1230 16 Taylor, S.L., King, J.W., and List, G.R (1993) Determination of Oil Content in Oilseeds by Analytical Supercritical Fluid Extraction, J Am Oil Chem Soc 70: 437–439... 437–439 17 Taylor, S.L., Eller, F.J., and King, J.W (1997) A Comparison of Oil and Fat Content in Oilseeds and Ground Beef Using Supercritical Fluid Extraction and Related Analytical Techniques, Food Res Int 30: 365 –370 18 Dionisi, F., Hug, B., Aeschlimann, J.M., and Houllemar, A (1999) Supercritical CO2 Extraction for Total Fat Analysis of Food Products, J Food Sci 64 : 61 2 61 5 19 Devittori, C., Gumy, D.,... 22.0 22 .6 42.3 42.3 42.7 34.7 34.3 34 .6 65.3 65 .7 65 .4 59.2 56. 8 45.4 2 .6 2 .6 2.5 95.8 95 .6 94.8 4.2 4.4 5.2 56. 9 54.4 42.9 0.5 2.2 11 .6 99.1 96. 1 78.8 0.9 3.9 21.2 (expressed as TAG), whereas the last extract ( . 18:1(n-7) Sample (%) Hull Extraction 1 22.5 42.3 34.7 65 .3 Extraction 2 22.0 42.3 34.3 65 .7 Extraction 3 22 .6 42.7 34 .6 65.4 Meat Extraction 1 59.2 2 .6 95.8 4.2 Extraction 2 56. 8 2 .6 95 .6 4.4 Extraction 3. 0.1 Flax 26 40.3 86. 7 96. 2 77.7 7.9 13.0 22.1 3.4 7.9 99.7 0.3 1 0 0.2 Solin 3 41.4 88.5 89 .6 87.8 0.9 10.9 12.0 9.2 1.5 99.4 0 .6 1.3 0.3 0 .6 Sunflower 20 45.4 87.9 94.5 76. 9 6. 1 11 .6 22.2 4.5 6. 1. Second >500 12 .6 9.8 6. 1 7.0 6. 4 0.0 355–500 26. 2 28.1 25.1 7.7 9.2 2.1 212–355 26. 7 27.4 29.2 17.8 23.8 16. 5 150–212 9.9 8.8 9 .6 17.9 18.1 19.8 75–150 9.9 7.0 5.2 16. 5 14.1 17.3 37–75 8.5 5.9 6. 8 33.1