TATE FLANT BOARD 
 
 E-491 September 1939 
 
 .!; United States Department of Agriculture 
 
 Bureau of Entomology and Plant Quarantine 
 
 A FIELD METHOD FOR THE CHEMICAL EVALUATION OF SPRAY 
 DEPOSITS RESULTING FROM THE APPLICATION OF INSECTICIDES 
 FOR CONTROL OF THE CODLING MOTH 
 
 By Jack E. Fahey and Harold W. Rusk 
 Division of Insecticide Investigations 
 
 INTRODUCTION 
 
 The methods of studying codling moth insecticides at Vincennes, Ind. , 
 are designed to fill the gap between the laboratory for testing toxicity 
 and the laboratory for testing insecticides on a large scale in the field. 
 Methods and techniques of analysis have been devised to permit detailed study 
 of residues on apples from the time of petal fall until harvest. To complete 
 the study of insecticidal spray residues, data from such chemical studies are 
 combined with similar data obtained by the Division of Fruit Insect Investi- 
 gations regarding the larvicidal value of residue deposits. The primary 
 purpose of these investigations is to develop and adapt new insecticides and 
 auxiliary materials for codling moth control, and this is accomplished through 
 study of the inherent qualities of residue deposits and the effect of growth 
 and weather on them. 
 
 Spray-residue analysis was first employed in the determination of 
 arsenical residues on harvested fruit in conjunction with studies of residue 
 removal. This type of analysis was also adopted later to show the value of 
 spreaders and nonarsenical insecticides in keeping the arsenical residue to 
 a low figure. Chemical analysis of spray residues was next employed in the 
 toxicity-testing laboratories, to measure the deposits produced by various 
 spray mixtures. It was then but a short step from the toxicity-testing 
 laboratory to the laboratory for field tests. For several years the study 
 of insecticidal spray residues encountered in codling moth control has been 
 an integral part of the plan of work of most laboratories concerned with 
 that insect. Basically these studies are similar in that residue deposits 
 are determined on samples taken at intervals during the growing season to 
 show the effect of multiple spray applications and weather on spray-residue 
 deposits. Such information has been particularly valuable in the planning 
 of spray treatments and in the comparison of different insecticides and 
 auxiliary spray materials. 
 
 The methods employed at the various experiment stations for studying 
 and reporting spray-residue deposits have not been standardized. As a re- 
 sult, direct comparison of data obtained at two stations employing different 
 
-2- 
 
 methods of study is impossible. Obviously it would be impossible here to 
 discuss the various methods employed at other experiment stations, but a com- 
 plete description will be given of the method of study used at the Vincennes 
 station, and certain practices at other stations which differ radically from 
 those employed at Vincennes will be indicated. 
 
 METHODS OF STUDY 
 
 In a chemical study of insecticidal spray residues the comparison 
 of results is actually the comparison of chemical analyses. Therefore such 
 comparisons involve all the errors of the analytical method plus the error 
 of sampling. To make a valid comparison of residues these errors must be 
 recognized and the magnitude of each estimated. Through the method of 
 sampling adopted attempts have been made to obtain a sample that would be 
 representative of the mean residue deposit within the plot studied. To 
 accomplish this all parts of the plot must be adequately represented in the 
 sample selected from that plot. A chemical method employed for the study 
 of spray residues must be suited to the type of material analyzed and the 
 range of spray residue encountered. In addition, it should be rapid to permit 
 frequent and numerous samplings, it should permit a precise reproduction of 
 results, and it should be reasonably accurate. Stress is placed on the 
 rapidity of the method because, to minimize the errors of sampling and to 
 permit a continuous study of the residues, any such study necessitates the 
 handling of a large number of samples. 
 
 Sampling Methods 
 
 Before attempting to discuss the methods of sampling employed it is 
 important that the problem be outlined. In the first place, spray residues 
 are not singular in any sense of the word. They represent the accumulated 
 deposits of a varying number of sprays; they are deposited on two or more 
 trees; they are not equally well deposited on all fruits of a single tree; 
 and they are not uniformly deposited on all the surface of a single fruit. 
 Furthermore, the residue on each fruit is affected by the rate of growth 
 of that fruit and the amount of weathering to which that fruit is exposed. 
 Thus the methods of study must consider all these variables and measure as 
 many of them as possible. To determine the value of each application of a 
 spray the studies must cover the entire period beginning with the initial 
 application of the insecticide and continuing until harvest. A represent- 
 ative number of trees must be included in each sample, and each sample must 
 adequately represent all sections of each tree. The effect of elements 
 such as weather and growth can be determined by frequent sampling of the 
 plots studied. The first sampling methods used were the outgrowth of pre- 
 vious experience and inspection, but later it was possible to investigate the 
 variations within plots and thus to modify the methods in accordance with the 
 facts as found. 
 
 There are three principal sources of variation in the amount of spray 
 deposit on apples within a plot of trees; first of these is the inherent 
 variation between trees due to their shape, their general structure, and the 
 
-3- 
 
 relative ease with which they can be sprayed; second, is variation within a 
 tree due to the position of the fruit upon it, that , is whether it is located 
 in the top or bottom, in the outer "shell" or on an inner branch, and whether 
 it is protected or unprotected by foliage; third, is variation of deposit 
 between different areas of a fruit, such as the stem, cheek, and calyx. Al- 
 though the variation in deposit on different sections of an individual fruit 
 is recognized, the method of study followed is based on samples in which the 
 smallest unit recognized is a single fruit. This is necessary in order that 
 the study of residue deposits may begin with the first spray application, at 
 which time it would be impossible to divide the fruit into its several parts 
 for analysis. To study the variation in deposit between trees and within 
 trees, analyses of individual fruits from trees have been made in plots on 
 which lead arsenate, fixed nicotine, and phenothiazine insecticides were 
 used. 
 
 Figure 1 represents a typical apple tree in a southern Indiana 
 commercial orchard and illustrates the way in v/hich the tree is divided 
 into segments, or strata, for sampling. Special attention is directed to the 
 fact that the lower branches nearly reach the ground, and at harvest it is 
 not uncommon to find numerous heavily laden branches resting on the ground. 
 The trees are arbitrarily considered as consisting of three horizontal sec- 
 tions, or strata, of equal height which by count contain approximately 30 
 percent of the fruit in each of the top and bottom strata and 40 percent in 
 the middle stratum. Varietal differences have been found to influence the 
 shape of the tree but have little effect on the distribution of the crop. 
 
 To determine the distribution of deposit on individual fruits, between 
 sections of trees, and between the trees of a plot, samples consisting of 
 single fruits have been analyzed. In these studies 105 samples were taken 
 from each plot of 5 trees, 21 fruits were selected from each tree, and 7 
 fruits were taken from each stratum of each tree. Several studies of this 
 nature have been made in plots sprayed with lead arsenate, those sprayed 
 with phenothiazine, and those sprayed v/ith nicotine-bentonite. These studies 
 were conducted over a period of 3 years and at different times during the 
 growing season. In table 1 a summarization of data from a study of plots 
 sprayed v/ith the three insecticides is presented. The deposits of pheno- 
 thiazine and nicotine are reported for plots sprayed with those materials 
 while the deposits from lead arsenate sprays are reported as the element 
 lead. These figures are accurate only for the specific study which they 
 represent, but such studies have been repeated a sufficient number of times 
 to show that the general relationships are the same regardless of the variety 
 of fruit used or the time of the growing season. The top stratum of the trees 
 always carries a significantly lighter deposit than the middle or bottom 
 stratum. Although the bottom stratum generally carries a heavier resudue 
 deposit than the middle stratum, the difference is not great and it is ques- 
 tionable whether it could be considered significant in all studies, especial- 
 ly where phenothiazine residues are concerned. The deposit on individual 
 fruits differs greatly with the insecticide under investigation, the more 
 durable lead arsenate deposit being much less variable than the phenothiazine 
 or nicotine deposits. 
 
^4- 
 
 Table 1. — A summary of the distribution of residue deposits of lead, 
 phenothiazine, and nicotine determined on individual fruits 
 from five-tree plots sprayed with lead arsenate, pheno- 
 thiazine, and nicotine-bentonite, respectively 
 
 R esidue (in mmg. per sq. cm. ) 
 
 Lead 
 
 Section of tree: 
 
 
 
 Top stratum 
 
 Maximum 
 
 28.0 
 
 
 Minimum 
 
 13.8 
 
 
 Average 
 
 19.9 
 
 Middle stratum 
 
 Maximum 
 
 52.1 
 
 
 Minimum 
 
 15.5 
 
 
 Average 
 
 31.2 
 
 Bottom stratum 
 
 Maximum 
 
 44.5 
 
 
 Minimum 
 
 22.5 
 
 
 Average 
 
 32.2 
 
 Entire tree 
 
 Maximum 
 
 52.1 
 
 
 Minimum 
 
 13.8 
 
 
 Average 
 
 28.1 
 
 lothiazine 
 
 Nicotine 
 
 36.5 
 
 4.9 
 
 7.7 
 
 .9 
 
 17.7 
 
 2.3 
 
 54.9 
 
 12.2 
 
 17.8 
 
 2.8 
 
 30.3 
 
 7.6 
 
 49.3 
 
 15.3 
 
 15.4 
 
 3.8 
 
 30.2 
 
 10.3 
 
 54.9 
 
 15.3 
 
 7.7 
 
 .9 
 
 26.1 
 
 6.7 
 
 Statistical analysis of the three sets of data from which table 1 
 was constructed and of similar data from other studies has shown the intra- 
 tree variation to be significant at odds of 99 to 1 in every study. On the 
 other hand the intertree variation was significant at odds of 99 to 1 only 
 when lead arsenate residues were studied. 7i/hen nicotine-bentonite and pheno- 
 thiazine deposits were studied, the intertree variation was usually sig- 
 nificant at odds of 19 to 1 or greater. By these studies two of the sources 
 of variation in an experimental plot have been measured and the intratree 
 variation has been found to be greater and more consistent than the intertree 
 variation. 
 
 For further development of a sampling method it is necessary to know 
 the size of sample (i. e., the number of fruits) required in order that 
 reasonable differences in deposit will be significant. To obtain these 
 figures the coefficient of variation has been calculated for the three sets 
 of data summarized in table 1 and other similar studies. The coefficients 
 of variation determined from the standard deviations of the studies described 
 above ranged from 22.5 to 37.7 percent, depending on the insecticide studied 
 and the time of the growing season at which the study was made. These data 
 indicate that differences of 10 percent of the common mean of samples com- 
 pared would be significant at odds of 19 to 1 when samples of from 60 to 90 
 fruits were used. 
 
 With these points in mind, an ideal sampling method should include all 
 tre3s of a plot, all strata of each tree, and, in order that differences of 
 
-5- 
 
 10 percent of the common mean would be significant at odds of 19 to 1, should 
 consist of 60 to 90 apples. To lessen the difference required for signifi- 
 cance or to increase the odds of significance it would be necessary to in- 
 crease the size of the sample. A practical method of sampling, however, must 
 consider the limitations of chemical methods, laboratory equipment, and the 
 personnel to handle the samples. 
 
 Samples for chemical analysis employed at Vincennes consist of 50 
 fruits divided into 25-fruit duplicates. Each 25-fruit duplicate is repre- 
 sentative of a single tree and is divided between the three strata of the 
 tree according to the normal crop distribution, that is, 7 fruits from the 
 top, 10 from the middle, and 8 from the bottom. Equal numbers of fruit 
 could be used without introducing any great error, but when a large number 
 of samples are taken during the growing season such a practice would inter- 
 fere with the normal crop distribution on the tree. Where studies are to be 
 conducted throughout the season five-tree or larger plots are generally 
 employed. All trees of a plot are sprayed according to a prearranged sched- 
 ule throughout the growing season. Two trees in each plot are used to pro- 
 vide samples, one of the pair being dropped and a new one substituted after 
 each spray so that all trees are sampled during the course of the season. 
 Thus each sample of 50 fruits represents all sections of 2 trees. The sample 
 has been divided between the different sections of the trees because these 
 show a greater variation than that between trees of a plot, and 2 trees of 
 the plot have been considered as being representative of that plot. In sam- 
 pling all sections of the tree the method differs from that employed by some 
 other laboratories, which confine their sampling to the lower sections. Sam- 
 ples taken from the lower sections of the trees or at shoulder height give 
 results higher than would actually occur on the entire tree. 
 
 Preparation of Samples 
 
 Two methods of recovery of residues from fruit are employed in sample 
 preparation. Inorganic residues, such as those of lead arsenate and calcium 
 arsenate, are removed from most fruits by peeling and digesting the peels, 
 but when fruit is small the entire fruit is digested. Organic residues 
 are usually removed from the surface of the fruit by selective solvents. 
 In preparation of the sample the residue from the entire fruit is recovered. 
 In this respect the method used at Vincennes differs from that preferred at 
 some experiment stations, where only the cheek surface or a part of the cheek 
 surface is used. By using the entire residue it is possible to begin studies 
 of deposits with the first spray application; on the other hand, methods in 
 which disks from the cheek surface, or the entire cheek surface are used are 
 not applicable until later in the growing season. Furthermore, the re- 
 moval of residues by the solvent method would be complicated if only the 
 cheek surface were employed. 
 
 Chemical Analysis 
 
 Several methods are employed in the analysis of spray residues of 
 arsenic, lead, nicotine, and phenothiazine, and each of these methods pos- 
 sesses features which make it especially useful for certain phg.ses of this 
 
-6- 
 
 study. Some of the methods employed in the analysis of each type of residue 
 are briefly outlined below, and the methods employed at the Vincennes station 
 are indicated. 
 
 Arsenical residues are determined by the bromate titration method. 
 This is used in preference to the Gutzeit method because it is both more 
 accurate and more precise. The bromate method is also better fitted to the 
 range of residues studied. Under ideal laboratory conditions both methods 
 require about the same amount of time for a complete analysis. 
 
 Lead residues are determined by the electrolytic method, dithizone 
 (diphenylthiocarbazone) being employed to isolate the lead before elec- 
 trolysis. This method is well adapted to the range of the residues studied 
 and is both more accurate and more precise than the colorimetric methods, 
 although the latter have some advantage in the time required for analysis. 
 
 Nicotine residues are determined by the colorimetric method employing 
 cyanogen bromide and beta-naphthylamine, developed by L. N. Markwood. Under 
 routine laboratory conditions this method is equal in precision and accuracy 
 to the gravimetric method, and is much more rapid since it does not require 
 steam distillation and ignition. 
 
 Phenothiazine residues are determined by the colorimetric method 
 involving treatment with bromine, developed by C. W. Murray. This method 
 has been found satisfactory for the analysis of chemically pure pheno- 
 thiazine; however, there is evidence that some of the decomposition products 
 of phenothiazine deposits will produce a color similar to that produced by 
 phenothiazine. It is therefore questionable whether the results of analysis 
 by this method give the true phenothiazine content of the residue deposit. 
 
 APPLICATION OF DATA 
 
 As will be seen from the foregoing discussion, relatively large 
 differences in residue deposits would be necessary to show significance 
 between any two samples. On the other hand, if the samples were increased 
 in size to permit small differences to show significance it would be impos- 
 sible to study more than three or four spray treatments throughout the 
 season. 
 
 During the 1938 season field studies at Vincennes covered a total of 
 55 spray plots; of these, 38 were compared in approximately 8 spray periods. 
 These plots were sampled twice during each spray period and each sample was 
 analyzed in duplicate. By employing the methods of sampling and analysis 
 outlined above 2 chemists and an unskilled laborer have been able to select 
 and analyze approximately 1,500 separate samples during a 130-day growing 
 season. In addition, fruit-growth records were obtained on three varieties 
 of fruit during the entire growing season. In the following discussion of 
 the application of data an attempt will be made to give a cross section of 
 the results of these studies by showing the various comparisons possible. 
 
-7- 
 
 Weathering of Residue Deposits 
 
 The simplest method of expressing residue deposits is in weight of 
 residue per fruit (micrograms per fruit). Such data are of value in the 
 study of the rate of weathering of spray deposits, since by disregarding the 
 distribution of residue on the fruit surface the effect of fruit growth on 
 samples taken on different days is nullified. This method is valid only 
 when it can be assumed that the fruits from the plots compared are relatively 
 uniform in size. Study of the weathering of spray residues is not a simple 
 matter, and because of the difference in the physical character of spray 
 deposits weathering may be expected to affect different deposits to different 
 degrees. In figure 2 are shown the residues, in micrograms per fruit, for 
 three totally different spray plots. Plot 1 received lead arsenate in nine 
 spray applications, but data from only the last eight are shown in this fig- 
 ure. Plot 2 received nicotine bentonite in eight spray applications and 
 plot 3 received phenothiazine in eight spray applications. Beginning at the 
 bottom of the chart it will be noted that in the case of plot 2 there is a 
 consistent trend to accumulate residue so that at the end of the season the 
 fruit retained approximately 20 percent of all residue deposited. The second 
 curve, that of plot 1, shows an even greater tendency for lead arsenate (in- 
 dicated by AS2O3 residue) to accumulate, and at the end of the season 66 
 percent of the total residue deposited was still present. Plot 3, treated 
 with phenothiazine, shows the greatest loss from weathering; between suc- 
 cessive spray applications this plot lost nearly all the residue from the 
 previous spray and at the end of the season retained only 11 percent of the 
 total residue deposited by eight sprays. As noted in the discussion of chem- 
 ical methods, it is probable that the results of analysis of phenothiazine 
 are inaccurate because some of the decomposition products will produce the 
 same color as phenothiazine under conditions of the analysis. Any such error 
 would tend to indicate greater deposits than those actually present on the 
 sample studied. 
 
 There is very little indication that rainfall alone is responsible 
 for the loss in residue by weathering. This is shown by the lack of cor- 
 relation between rainfall and loss of residue. It is quite probable that 
 brushing of the fruit by foliage movement is more important than rainfall 
 in removal of residue after the residue has once dried on the fruit. Heavy 
 rains immediately following a spray application, as occurred on August 5, 
 have a pronounced effect on phenothiazine residues and a measurable effect 
 on nicotine residues, as is indicated in the very small quantity of residues 
 deposited on this date. 
 
 Fruit-Growth Records 
 
 Fruit growth is an important part of any study of spray residues on 
 apples, and such data can be easily obtained from the weight and number of 
 fruits in samples analyzed. To obtain the area of any sample the apple 
 was considered as a sphere and the area calculated from its weight and its 
 specific gravity. 
 
 ^.-^'^^-^ 
 
-8- 
 
 In figure 3 a chart has been prepared to represent the relative sur- 
 face area of one fruit at different periods during the growing season. 
 These figures were obtained from growth records on Grimes Golden apples in 
 the Vincennes area covering 3 years. The figures in the column marked "days" 
 at the extreme left represent the days from petal fall (calyx-spray date). 
 The middle column on the' left of the figure identifies the spray which would 
 be applied on that day and for that size of fruit. The column on the left 
 next to the figure gives the average daily percentage increase in surface 
 area between days and sprays. The figures within the several squares repre- 
 senting the size of the fruit are the average surface areas of one fruit in 
 square centimeters. 
 
 Three years' records of Grimes Golden fruit growth show an average 
 daily increase of 1 square centimeter per apple per day, At the blossom 
 period this will amount to as m.uch as 100 percent increase in surface area 
 per day or, as indicated by this figure, an average daily increase of 22.3 
 percent per day between the 7-day spray and the first cover spray. Such a 
 rapid development of area of fruit rapidly decreases the surface concentra- 
 tion of residue deposits. As the size of the fruit increases the relative 
 rate of growth is materially reduced; consequently, during the second-brood 
 period the daily increase in area amounts to only a little more than 2 per- 
 cent per day. 
 
 Thus these data show the rate of growth of fruit and its effect on 
 residue deposit. It is interesting that normal spray applications during 
 the first 45 days of the growing season (approximately the first-brood 
 period) do not deposit sufficient residue to compensate for the growth of the 
 fruit. During the second-brood period the normal spray application will 
 deposit sufficient residue to compensate for growth and will accumulate an 
 excess deposit. 
 
 Residue-Area Relationship 
 
 The control of insects by spray deposits depends on the concentration 
 of residue on the fruit surface. To obtain such a figure the residue per 
 fruit is divided by the area per fruit. If the residue were expressed in 
 micrograms and the area in square centimeters this would give a value in 
 micrograms per square centimeter. In figure 4 the data obtained from two 
 nicotine spray plots employed during the 1938 season are graphically illus- 
 trated. These data are reported in micrograms of nicotine per square centi- 
 meter of fruit surface to permit a direct comparison of the average con- 
 centration of deposit on the surface of fruit from two or more plots. Such 
 data include the effect of both weathering and fruit growth on residue de- 
 posit. The figures represent the results of 18 analyses of 50-fruit samples 
 taken from each plot on the days indicated. Eight sprays were applied to 
 each plot. It may be noted that the difference between these two plots at 
 any one date is not great and in most cases would not be significant at odds 
 of more than 19 to 1. On the otherhand, the data consistently favor one 
 plot and there can be little question as to the significance of the differ- 
 ence that exists between these tv/o treatments. These data indicate the man- 
 ner in which small differences can assume significance when a study is con- 
 tinued through an entire growing season. 
 
-9- 
 
 There are two other characteristics of spray residues which are 
 well illustrated in this figure. (1) Early spray applications (that is, 
 spray on small fruit which has not lost its pubescence) deposit greater 
 residues per unit area than late spray applications. Whether this is due to 
 the character of the fruit surface or to the ease of completely covering 
 small fruits has not been definitely determined. (2) The slowing down of the 
 relative rate of increase of fruit surface late in the spray season makes 
 it appear that deposits from late cover sprays are more persistent than 
 deposits from early sprays. 
 
 Such data as are presented in this figure are valuable for direct 
 comparison of spray treatments. The method permits a comparison between 
 deposits at any one period or, when considered as a whole, throughout the 
 season. The combined effect of weather and fruit growth on residue deposit 
 is also shown. 
 
Digitized by the Internet Archive 
 in 2013 
 
 http://archive.org/details/fieldnnethodforchOOunit 
 
MIDDLE 40 *y. 
 
 BOTTOM 30®/ 
 
 Figure !•— Crop distribution on a typical apple tree 
 in a coomeroial orohard in Indiana. 
 
 
Figure 2. —Effect of weathering on spray deposit, 
 

 1 
 
 
 
 
 
 
 
 
 
 
 « 
 
 
 
 Ol 
 
 
 
 
 
 
 
 
 
 
 
 • 
 
 2 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 o 
 
 
 
 
 
 
 
 
 
 
 
 z 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 1- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 rs 
 
 
 
 
 
 
 
 
 • 
 
 
 
 cc 
 
 
 
 
 
 
 
 
 
 
 o 
 
 u. 
 
 
 
 
 
 
 
 
 
 
 <o 
 
 T< 
 
 
 
 
 
 
 
 
 
 t*N. 
 
 
 u. 
 
 o 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 < 
 
 
 
 
 
 
 
 
 Oi 
 
 
 
 lU 
 
 
 
 
 
 
 
 •1 
 
 
 
 
 Of 
 
 
 
 
 
 
 
 s. 
 
 
 
 
 < 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 lo 
 
 
 
 
 
 
 
 
 
 
 lO 
 
 
 
 
 
 
 
 
 
 
 
 <n 
 
 
 
 
 
 
 
 
 
 
 
 CO 
 
 
 
 
 
 
 
 
 
 
 1; 
 
 
 
 
 
 
 
 
 
 
 
 c\i 
 
 
 
 
 
 
 
 
 
 
 
 <sj 
 
 
 
 
 
 
 
 
 
 
 N 
 
 
 
 
 
 
 
 
 
 
 
 T 
 
 
 
 
 
 
 
 
 
 
 <0 
 
 
 
 
 
 
 
 
 
 
 
 
 
 vS 
 
 
 
 
 
 
 
 
 
 
 <0 
 
 
 
 
 
 
 
 
 
 
 
 N 
 
 
 
 
 
 
 
 
 
 
 o 
 
 (0 
 
 s 
 
 (d 
 o 
 
 O 
 
 +> 
 O 
 0) 
 
 <H 
 
 4) 
 
 •H 
 
 >- 
 
 or. 
 
 Q. 
 
 to 
 
 10 
 
 CO 
 fvi 
 
 I 
 
 Ifl 
 
 O 
 
 z 
 
 CM 
 
 csj 
 
 lO 
 
 Q 
 
 
 X 
 
 
 I- 
 
 «q 
 
 CO 
 
 c5 
 
 t^J 
 
 3 
 
 3 
 00 
 
 3 
 
 o 
 
 P 
 
 ^ 
 
to 
 +^ 
 o 
 
 o 
 
 +> 
 O 
 
 oj o 
 
 o -P 
 0) 
 
 ;3 I 
 
 "-* fl 
 ra ^ 
 (j> +> 
 fH o 
 o 
 
 o rt 
 
 ^ 
 +> 
 
 o 
 
 tn 
 
 •H 5 
 
 u 
 
 P< (D 
 S !>. 
 
 cd 
 
 1 ^ 
 
 'St- 
 0) 
 
 (i« 
 
UNIVERSITY OF FLORIDA