Org. Gechem. Vl. 17, N. 5, pp. 635-648, 1991 0146-6380/91 $3.00 + 0.00 Printed in Great Britain. All fights reserved Cpyright 1991 Pergamn Press plc Statistical evaluatin f the elemental cmpsitin f humic substances JAMES A. RIC~* and PATRICK MACCARTHY Department f Chemistry and Gechemistry, Clrad Schl f Mines, Glden, CO 80401, U.S.A. (Received 19 April 1990; accepted 7 December 1990) Abstract--Elemental data (C, H, O, N, S, atmic H/C and O/C ratis) fr humic acids (410 samples), fulvic acids (214 samples) and humin (26 samples), islated frm envirnments all ver the wrld, were cmpiled frm the literature. These data were analyzed statistically using the mean, median, mde, range, standard deviatin and t-test. The large data base allws statistically significant differences between varius humic substances t be established. Prir t sme f the statistical evaluatins, the humic substances were gruped accrding t surce---sil, freshwater, marine r peat. The average elemental cmpsitins fr humic acid, fulvic acid and humin, in tw, and als when segregated by surce, are presented. Standard deviatins fr carbn cntents are remarkably small suggesting that perhaps an ptimum cmpsitin exists fr humic substances in nature. The evaluatin shws that humic acids in tw are significantly (P < 0.0005) different frm fulvic acids in tte with respect t C, N, and O cntents and atmic H/C ratis. Fulvic acid in tte is significantly (P < 0.0005) different frm humin/n tte in terms f C and O cntents. When segregated by surce, sme significant (P < 0.0005) differences between humic acids islated frm freshwater, marine, and sil envirnments are evident; similarly, significant differences are fund between fulvic acids frm freshwater and sil surces. Three van Krevelen diagrams based n: (1) the 650 samples f humic acid, fulvic acid and humin; (2) humic acids segregated by surce, and; (3) fulvic acids segregated by surce, are discussed. Key wrds--humic acid, fulvic acid, humin, elemental cmpsitin, statistical analysis INTRODUCTION One f the mst fundamental characteristics f a chemical substance is its elemental cmpsitin. Determining the elemental cmpsitin (r empirical frmula) f a discrete chemical cmpund is a first step t btaining its mlecular frmula; this, in turn, is critical fr determining its structural frmula. Hwever, in the case f nn-stichimetric materials such as cal, kergen and humic substances, the cncept f even a net mlecular frmula is extremely limited, and a unique, structural frmula des nt exist fr these materials. Nevertheless, elemental analysis is a useful tl fr characterizing nn-stichimetric mixtures such as humic substances. The applicatins f elemental analysis in the study f humic substances have recently been reviewed by Steelink (1985). Because humic substances are nn-stichimetric materials they must be characterized in terms f their average prperties. Accrdingly, it is apprpriate t cnduct a statistical evaluatin f these prperties in rder t establish the statistically valid differences and trends in these materials. Elemental cmpsitin is ne such average prperty and it is statistically evaluated in this paper. *Present address: Department f Chemistry, Suth Dakta State University, Brkings, SD 57007, U.S.A. van Krevelen (1961) develped a graphical methd t study the calificatin prcess in which the atmic hydrgen/carbn (H/C) rati is pltted as a functin f the atmic xygen/carbn (O/C) rati. This type f plt, nw generally knwn as a van Krevelen diagram, is ften used fr the classificatin f cals and kergens. A frequent applicatin f the van Krevelen diagram is t illustrate the changes in elemental cmpsitins that ccur during the alteratin f rganic gechemicals in a gelgic envirnment; e.g. H/C and O/C ratis have been used t fllw the effects f diagenesis n humic substances (Huc and Durand, 1977; Reuter and Perdue, 1984). van Krevelen diagrams have als been used by varius wrkers t illustrate cmpsitinal differences between humic acids and fulvic acids, and als t shw variatins in humic substances as a functin f surce. Fr example, Kuwatsuka et al. (1978) used a van Krevelen diagram t cmpare the elemental cmpsitins f sil humic and fulvic acids, cals, plant tissues and varius classes f rganic cmpunds. Visser (1983) emplyed a van Krevelen diagram t cmpare fulvic and humic acids frm aquatic and terrestrial surces. The magnitude f the H/C rati has als been used t indicate the degree f armaticity r unsaturatin (a small value) r aliphaticity (a large value) f a substance (van Krevelen, 1961). Perdue (1984) has pinted ut that the ttal unsaturatin f a humic 635
636 JAMES A. RJc'~ and P^rmCK M^CC~Th'V material cannt be btained slely frm the H/C rati; in additin t unsaturated frms f carbn the H/C rati is als a functin f unsaturatin present in functinal grups, primarily carbxyl and carbnyl grups, with lesser cntributins frm ther miscellaneus frms f unsaturatin. If H/C ratis are calculated fr the 21 humic material samples in the study f Perdue (1984) and cmpared t the armatic carbn cntents crrected fr the varius frms f nncarbn unsaturatin it is seen that, thugh the actual numbers differ, samples which exhibit a high armatic carbn cntent als exhibit a small H/C rati and vice versa. The lne exceptin is a spdsl fulvic acid with high ttal acidity (12mequiv/g), a lw crrected armatic carbn cntent but a mderate H/C rati (0.85). The H/C rati thus appears t be a qualitatively useful parameter fr cmparing the armaticities f humic materials. T date, the number f humic samples pltted n a single van Krevelen diagram has been relatively small. The value f any such investigatin wuld be enhanced by enlarging the data base and by using humic substances frm a wider variety f surce envirnments. In additin, when a large data set is emplyed ne can justifiably apply statistical methds t quantify the relatinships between the varius grups f humic substances by establishing the statistically significant differences. Fr this study we cmpiled frm the published literature an extensive data base cnsisting f the elemental cntents f 650 samples f humic substances. The bjectives f this paper are: (1) t apply basic statistical methds t a very large data base f elemental cntents in rder t quantify the relatinships between varius classes f humic substances; and, (2) t study van Krevelen diagrams fr humic substances prepared frm this large data base. Elements studied DATA AND DATA ANALYSIS Elemental cntents (C, H, N, S, and O) f humic acids, fulvic acids and humins were cmpiled frm the literature. Elemental cntents are generally reprted n a weight/weight percentage basis and that is the frm used in this paper. All elemental cntents are expressed n an ash-free basis, and, where necessary, the ash-flee values were calculated by the present authrs using the published ash-cntents. Atmic H/C and atmic O/C ratis were cmputed fr all samples and were pltted n a van Krevelen diagram. Fr this study, the samples were classified int humic acid, fulvic acid and humin n the basis f the cnventinal definitins f these fractins and based n the descriptins f the samples presented in the riginal publicatins. The samples were further classified as sil, freshwater, marine, peat and cal humic materials, depending n their surces. The samples described as marine were btained almst exclusively frm marine sediments; nly tw fulvic acids and tw humic acids in this data set represent disslved marine materials. Dpplerite, lignite and lenardite humic substances were included with the cal samples. The humic acid set cnsisted f 410 samples, the fulvic acid set had 214 samples, and there were 26 samples f humin. These samples had been islated frm lcatins all ver the wrld. Because f the large number f analyses cmpiled, n single authr r publicatin dminates the cntributins t the verall data sets. The elemental data fr each sample cnsisted f carbn, hydrgen, and xygen cntents in all cases, nitrgen cntents fr all but tw samples, and sulfur cntents fr apprx. 30% f the samples. Only in a few cases had xygen been determined directly; generally it had been btained by difference. Statistical analysis The mean, ($), was cmputed using the fllwing equatin: tl ~x~ =~=i (1) n where n is the ttal number f samples in a set and x~ represents the individual values fr a particular parameter. The median (the value abut which all ther values are equally distributed) and the mde (the mst frequently ccurring range f values) were als btained frm the data set. The mde was nt determined fr data sets f <20 samples nr is it presented fr small data sets which exhibited 2 r mre intervals f the same frequency. The mean, median and mde are measures f the central tendency fr that parameter and are the same fr nrmally distributed data. The range, as presented, gives the lw and high values fr a parameter in the data set. The standard deviatin (s) was calculated using equatin (2): 1-- n 2--11/2 s= L. -j. (2) The range is a measure f the dispersin f the data, and standard deviatin is a measure f the dispersin f the data arund the mean. The t-test used here evaluates the prbability that the mean value f a particular parameter exhibited by tw data sets (.~ and $2) culd be bserved within the same ppulatin. The t-statistic is btained using equatin (3): t=[n2(~ l)j I'`2 (3)
Statistical evaluatin f the elemental cmpsitin f humic substances 637 where Q is defined by: Q=n, (x~, - x~,) nl r rt [ "l i Ln2J J --i~=l(xl--xei)l~j rntll/212.] (4) and nt and n 2 are the number f measurements in the smaller and larger data sets, and xl, and x2, are the individual values in the smaller and larger data sets, respectively (Crw et al, 1960). This study emplyed a ne-sided t-test. Chi-squared (X 2) statistics are used t assess the nrmality f the distributin f values in the data set fr a particular parameter. This infrmatin is necessary because the t-test is parametric, i.e. it perates n the assumptin that the data exhibit a nrmal distributin. Cnsequently, the data in an untransfrmed state and lgm-transfrmed state were evaluated using the fllwing equatin (Crw et al., 1960): X2= (t/i- ei) 2 (5) /.d i=t where r is the number f categries, r/~ are the bserved frequencies and ei are the theretical frequencies. The frm f the data which exhibited, r mst-clsely apprximated, a nrmal distributin was then used in the t-test. In mst instances the data in the untransfrmed state exhibited, r mst-clsely apprximated, a nrmal distributin. The use f data transfrmatins in statistical analysis is discussed by Bartlett (1947). Figure 1 gives a series f histgrams which illustrate the distributin f the data which was judged t exhibit, r mst clsely apprximate, a nrmal distributin fr three f the data sets used in ei this study. A visual assessment f nrmality can be made fr the larger data sets (e.g. all humic acids) but t d s becmes increasingly difficult as the number f samples decreases. Chi-squared statistics are usually a mre sensitive indicatr f the distributin f the data (Crw et al, 1960) and fr this reasn they were used t evaluate the nrmality f the data when pssible. Chi-squared statistics were nt calculated fr data sets cntaining < 50 samples. When an assessment f nrmality was needed fr small data sets it was made by inspectin f the histgrams. The degrees f freedm (d.f.) were estimated using Satterthwaite's (1946) apprximatin. This methd estimates the degrees f freedm in situatins where it is assumed r knwn that sl :/: s2. The degrees f freedm are btained using: s~+ r s;t d.f. = Lnl n2j (6) Cmputer prgrams nl m n2 The FORTRAN prgram PHIST (Klusman, 1981) was used t calculate the statistical quantities described in equatins (1), (2), and (5). The prgram rders the values in a data set frm lwest t highest which facilitates the determinatin, by inspectin, f the median and range. PHIST als plts the distributin f the data as nrmal standard deviate histgrams frm which the mde may be btained by inspectin. Tests f significance [equatin (3)] were perfrmed by TTEST (Klusman et al., 1980), a FORTRAN prgram fr calculating the mean and t-statistic fr tw sample sets in which it is knwn, r assumed, that S 1 S 2. C H N S 0 O/C H/C,(::,,,,II,,,,,,,,dh,,,,,,,,i,l,,,,,...,,,.,,,,....,,,,,.....,,tl,,,....,,,,It,,... ALL humic acids (b)...,,,,...,,,,h,l,,...,,111,1,...,,,,l,,...,,llll,,,,... Sil humic acids,,,i II Ih,,...,,,,I I,,,... (c) A... hi,,i...,tllh..... I,IIIli...,,,..,...... till...,,tl... h,,lm,.,..,. Marine humic acids Fig. l. Elemental and atmic rati histgrams fr (a) unsegregated humic acids, (b) sil humic acids and (c) marine humic acids (based n data in Tables l and 3). All data except the S values are untransfrmed; the S data are logl-transfrmed.
638 JAMES A. PaCE and P^TPdCK MACCARTHY Limitatins and assumptins in this study This study is based n the analysis f experimental data btained frm the literature; as such, the quality f the analytical methdlgy per se in the individual publicatins frm which the data were btained cannt be directly evaluated. Hwever, even thugh the elemental analysis f humic substances is nt withut its prblems (Huffman and Stuber, 1985), carbn and hydrgen determinatins can generally be carried ut with a high degree f accuracy and precisin. It is extremely rare fr the results f replicate determinatins t be reprted in the elemental analysis f humic substances and thus ne cannt evaluate the analytical precisin pertaining t the individual samples; Huffman and Stuber (1985) in a unique study d, hwever, reprt n the precisin assciated with the elemental analysis f humic substances. Fr the present study, it must be assumed that the analytical data are accurate and that the dispersin in the data results slely frm genuine differences between the samples. Generally, xygen cntent is nt determined directly, but is btained by subtracting the sum f the ther elemental cntents, plus the ash cntent, frm 100%. As pinted ut by Huffman and Stuber (1985) this prcedure is subject t tw serius limitatins: (1) the xygen cntent determined in this manner includes the sum f all errrs in the ther elemental determinatins, and (2) the ash may cntain elements which have been previusly determined, s that subtractin f the ash cntent, in effect, subtracts thse elements a secnd time. In sme studies xygen has been determined directly (Huffman and Stuber, 1985; Malclm and MacCarthy, 1986) and generally if the six elements C, H, N, O, S and P are determined, the summatin f elemental cntents, n an ash-free basis, cmes clse t 100%, indicating that the rganic matter in humic substances is cnstituted almst entirely frm these six elements. Hwever, in the case f sme high-ash samples, direct determinatin f xygen can result in grssly errneus values (Malclm and MacCarthy, 1986; MacCarthy and Malclm, 1989). Anther prblem in the determinatin f xygen by difference is that, generally, nly C, H and N cntents are measured directly, and, cnsequently, the value calculated fr xygen by subtractin includes ther elements such as sulfur and phsphrus. Hwever, as will becme evident frm the data cmpiled fr this study, sulfur cntent can represent as much as 3% in sme samples f humic substances. In additin, cmparatively cmplete analyses f humic substances (Thurman and Malclm, 198 I; Huffman and Stuber, 1985) have shwn that even phsphrus may cnstitute a measurable fractin f humic substances. Fr this study the published xygen cntents must be accepted at face value fr use in the statistical calculatins, but the abve limitatins relating t the xygen values shuld be brne in mind. The nntrivial prblems assciated with misture, and misture determinatin, in humic substances are discussed by Huffman and Stuber (1985), wh pint ut that even sme cmmercial labratries experience difficulty with this prblem. The presence f misture can lead t tw types f errrs: (1) the hydrgen and xygen cntents f the misture may be incrrectly attributed t hydrgen and xygen in the rganic matter, and (2) the presence f misture can lead t an incrrect value fr the ttal rganic matter cntent unless prperly crrected fr. Again, in this type f study, we cannt assess the extent f these prblems, but must assume that the misture cntents were prperly determined and crrected fr in the surce publicatins, which mst prbably will nt be a crrect assumptin in sme cases. As a result f these cnsideratins, the carbn data may be the least susceptible t errr, and cnclusins based n the distributin f the element may be the mst reliable. Overall, it is reasnable t assume, based n the results which fllw, that the majrity f the elemental determinatins were carefully cnducted and that grssly errneus values represent nly a small fractin f the ttal sample ppulatin and will nt seriusly influence the statistically-based cnclusins f this paper. RESULTS AND DISCUSSION Humic substances in general (humic and fulvic acids) (1) General trends in elemental cmpsitin fr humic substances. The mean, median, mde, range, standard deviatin and X2-statistics fr elemental cmpsitins f the humic acids, fulvic acids and humins are given in Table 1. The number f significant figures fr each entry in this and subsequent tables is based n the values reprted in the surce publicatins. It shuld be recgnized that it is the mean r average values that are being cmpared in this paper, and that the ranges fr elemental cntents f varius humic fractins, as taken frm the literature, actually display cnsiderable verlap (cf. Table 1). Cnsequently, individual samples may deviate frm the general trends that are reprted in this paper. In all data sets the range is defined by the utlier, r extreme, values. Cnsequently, sme f the values specifying the range in Table 1 (and subsequent tables) may appear highly anmalus; e.g. a sil humic acid utlier has a H/C rati f 0.08, and a sil fulvic acid utlier has a H/C rati f 2.13 (Table 1). Hwever, these are extreme, and pssibly spurius, values and their ccurrence is t be expected in a large data set, particularly f the type under discussin here. These extreme values are bviusly nt representative f the data set as a whle, as evident frm cmparing them with the crrespnding mean values and standard deviatins. Table 1 incrprates humic substances frm all envirnments (sil, freshwater,
Table I. Mean, median, mde, range, standard deviatin (SD) and 12-statistics fr elemental cmpsitins f unsegretated humic substances expressed as weight percent; O/C and H/C ratis are atmic percent. All values are n an ash-free basis r~ c H N S O O/C H/C Humic acid ~" Mean 55.1 5.0 3.5 1.8" 35.6 0.50 1. I 0 ~,. Median 55.7 5.0 3.3 1.1 35.1 0.47 1.10 Mde 55.2-56.4 (61) 4,7-5.0 (59) 3.1-3.4 (54) -- 34.1-35.5 (78) 0.47-0.50 (76) 1.10-1.16 (64) Range 37.18-75.76 1.64-11.68 0.50-10.54 0.1-8.3 7.93-56.6 0.08-1.20 0.08-1.85 ~" SD 5.0 (9.1) 1.1 (22.0) 1.5 (42.9) 1.6 (88.9) 5.8 (16.3) 0.13 (26.0) 0.25 (22.7) ---. Q Z2-statistic 111.8 (41.3) 88.8 (41.3) 45.3 (41.3) 22.8 (23.4) 97.0 (41.3) 196. i (41.33) 49.7 (41.3) N. f samples 410 410 410 160 410 410 410 Fu/v/c ac/d ~ Mean 46.2* 4.9 2.5* 1.2" 45.6 0.76 1.28* Median 45.6 4.7 2.3 0.9 46.3 0.76 1.21 D ~" Mdc 42.2-43.4 (25) 4.1-4.3 (31) 2.9-3.4 (32) 1.3-1.7 (13) 44.2-45.6 (25) 0.68~.72 (20) 1.04-1.12 (35) Range 35.1-75.4 0.43-7.2 0.45-8.16 0.10-3.6 16.9-55.8 0.17-1.19 0.77-2.13 = SD 5.4 (I 1.7) 1.0 (20.4) 1.6 (64.0) 1.2 (100.0) 5.5 (12.1) 0.16 (21.1) 0.31 (24.2) ~" ;t 2-statistic 39.4 (30.1) 40.9 (30.1) 51.9 (30.1) 6.6 (I 1. I) 31.9 (30.1) 36.8 (30.1) 34.3 (30.1) 8 N. f samples 214 214 214 71 214 214 214 [~ ~t Hum/n. Mean 56.1 5.5 3.7 0.4* 34.7 0.46 1.17 -~. Median 56.3 5.5 3.3 0.3 34.4 0.46 1.18 = Mde 56.1-56.8 (7) 4.0-4.5 (6) 2.4-2.7 (4) 0.2-0.6 (9) 33.9-34.7 (6) 0.454).46 (7) -- Range 48.29~1.60 4.2-7.28 0.90-6.00 0.1-0.9 28.80-45.12 0.374).61 0.82-1.72 ~r SD 2.6 (4.6) 1.0 (18.2) 1.3 (35.1) 0.3 (75.0) 3.4 (9.8) 0.06 (13.0) 0.24 (20.5) E,~ 2-statistic........ N. f samples 26 26 24 16 26 26 26 An asterisk (*) indicates that this parameter in the humic material exhibited r mst-clsely apprximated a lg-nrmal distributin; all ther values exhibited a nrmal distributin. Values in parentheses fllwing the mde are the number f samples in that interval, thse fllwing the standard deviatin are relative standard deviatins (RSD). 2-Statistics are prvided fr the mst nrmal frm f the data as indicated abve and Z 2 fr that sample size fllws in parentheses.
640 JAMES A. RICE and PATRICK MACCARTHY marine, peat, cal) and they are nt segregated by surce. In general, there is clse agreement between the mean, median and mde, cnsistent with the data being nrmally distributed. Even thugh the carbn cntents f the humic acids range frm 37.18 t 75.76% the standard deviatin fr thse samples is nly 5.0% [relative standard deviatin (RSD) = 9.1%]. Sixty-eight percent f the humic acids have carbn cntents in the interval 50.1-60.1% (+ 1 SD) and 96% f these samples have carbn cntents between 45.1 and 65.1% (+2 SD). Cnsidering that (1) humic acid is simply an peratinally-defined prduct, (2) the samples invlved in this study were btained by numerus variatins f the traditinal extractive technique, (3) that these ht/mic acids were extracted by many different experimentalists, and (4) that these humic acids represent a brad diversity f surce envirnments (sil, freshwater, marine, peat and cal), gegraphical lcatins and btanic precursrs, the abslute standard deviatin f nly 5.0% (9.1% RSD) is ntewrthy. In fact, it was precisely this recgnitin f suprisingly small standard deviatins, here and elsewhere in this study, that prmpted us t embark n a mre rigrus statistical evaluatin f the elemental cmpsitins f humic substances. This bservatin may suggest that an ptimum cmpsitin exists fr humic acids in nature. This, hwever, des nt imply the existence f mlecular structural unifrmity amng humic substances. The range and relative standard deviatin fr the carbn cntents ffulvic acids are smewhat larger than thse fr humic acids, and the crrespnding values fr humin are cnsiderably smaller (Table 1). Hwever, ne cannt speak as reliably f the humin data because nly 26 such samples were invlved in the study, and mre than half f these were frm peat; these limited data may nt be representative f humin in general. The mean values fr the H/C and O/C atmic ratis based n these large data sets are similar t thse previusly reprted fr humic acid and fulvic acid (Visser, 1983; Steelink, 1985) and fr humin (Stuermer et al, 1978) based n much smaller data sets. The H/C rati (Table 1) indicates that fulvic acids are, in general, mre aliphatic than humic acids. It is seen frm Table 1 that the ranges in reprted carbn cntents fr humic acids and fulvic acids are similar. Hwever, a single utlying datum can have an inrdinate influence n the range, but the mean carbn cntents fr humic and fulvic acids are cnsiderably different, displaying the net differences between these materials. The ranges fr ther elemental cntents are als rather large (Table 1). The relative standard deviatins fr hydrgen are mderately high, fr nitrgen are quite large and fr sulfur are very large (88.8% fr humic acid and 83.3% fr fulvic acid). The large standard deviatins fr nitrgen and sulfur may simply be a result f inherent variability f these Table 2. t-statistics, and their significance, used t cmpare unsegregnted humic acids (HA), fulvic acids (FA) and humins (HMN) frm Table I t-statistic t c C H N O O/C H/C HA-FA 3.29 19.66" 2.08 6.47* 21.41" 20.03* 7.67* HA-HMN 3.58 1.64 2. I 1 1.48 I. 17 2.51 1.43 FA-HMN 3.48 17.68" 2.80 2.76 12.54" 8.75* 1.51 t c Represents the critical t-value at P = 0.0005 fr each cmparisn and an asterisk (*) indicates a t-value significant at P < 0.0005. elements in humic substances, and may als be due t analytical errrs in the determinatin f these lesser cnstituents. On the basis f these data, and fr the reasns stated in the previus sectin, cnclusins based n bserved trends in the means may be mst reliable in the case f carbn. (2) Statistically significant differences in humic substances. Numerus trends can be discerned in the data f Table 1. The imprtance f such relatively large data sets is that they allw the statistically significant trends t be established by means f the t-test. Table 2 presents the t-statistics, calculated using equatin (3), which cmpare humic acid with fulvic acid, humic acid with humin, and fulvic acid with humin fr each f the elemental cntents and atmic ratis f the unsegregated samples described in Table 1. Due t the cmparatively small number f sulfur analyses, the statistical significance f trends bserved in the sulfur cntents are nt evaluated. The prbability, P, at which t reject the null hypthesis equating the means f a parameter exhibited by tw data sets has been set at P < 0.0005. This means that if the t-statistic fr a particular parameter exceeds the critical value (t) there is less than ne chance in 2000 that the means exhibited by thse tw data sets culd be bserved in the same ppulatin; the values that fall int this categry are labeled with an asterisk in Table 2. Many f the t-values in Table 2 are very much larger than the critical values, t; in these cases the null hypthesis is rejected at P-values that are cnsiderably less than 0.0005. Fulvic acid is statistically distinct frm humic acid n the basis f its C (lwer), N (lwer) and O (higher) cntents and its O/C rati (higher). Fulvic acid als exhibits a larger and statistically different H/C rati relative t humic acid. Because a larger H/C rati is indicative f a mre aliphatic character, this is cnsistent with fulvic acid being, in general, mre aliphatic than humic acid. The nly parameter fr which fulvic acid is nt significantly different frm humic acid in Table 2 is hydrgen cntent. Fulvic acid differs statistically frm humin with respect t its C (lwer) and O (higher) cntents and its higher O/C rati. The statistical cmparisn f humic acid and humin des nt shw these tw materials t be significantly different at P < 0.0005, and cannt discunt the pssibility that the parameter means exhibited by the respective data sets in this study might be bserved in the same ppulatin at the chsen
Table 3. Mean, median, mde, range, standard deviatin (SD) and X2-statistics fr elemental cmpsitins f humic acids frm different surces expressed as weight percent; O/C and H/C ratis are atmic percent. All values are n an ash-free basis C H N S 0 O/C H/C Sil humic acids Mean 55.4 4.8 3.6 0.8* 36.0 0.50* 1.04 Median 56.0 4.9 3.7 0.6 35.4 0.47 1.06 Mde 56.3-57.3 (38) 5.0 5.3 (32) 3.9-4.2 (30) 0.5-0.6 (14) 35.0-36.0 (34) 0.47-0.49 (44) 1.04-1,10 (38) Range 37.18-64.1 1.64-8.0 0.5007.00 0.1-4.88 27.1-51.98 0.33-0.97 0.08-1.77 SD 3.8 (6.9) 1.0 (20.8) 1.3(36. l) 0.6 (7.5) 3.7 (10.3) 0,09 (18.0) 0.25 (24.0) X 2-statistic 72.1 (30.1) 51.6 (30.1) 39.7 (30.1) 9.2 (9.5) 55.7 (30.1) 71,7 (30.1) 41.2 (30.1) N. f samples 215 215 215 67 215 215 215 Freshwater humic acids Mean 51.2 4.7 2.6* 1.9* 40.4 0.60 1.12 Median 5 I. 7 4.7 2.1 1.1 40.8 0.60 1.12 Mde 51.2-51.9(7) 4.7--4.9(8) 1.9-2.2(12) 0.6-2.0(8) 40.4-41.4(12) 0.56-0.58(10) 1.12-1.16(9) Range 43.75-56,00 3.5-6.54 0.63-7.97 0.35-4.31 30.86-48.2 0.42-0.80 0,79-1,69 SD 3.0 (5.9) 0.6 (12.8) 1.6 (61.5) 1.4 (73.7) 3.8 (9.4) 0.08 (13.3) 0.17 (15.2) X2-statistic 1.3 (7.8) 2.0 (7.8) 5.6 (7.8) 5.6 (7.8) 3.5 (7.8) 3.5 (7.8) N. f samples 56 56 56 13 56 56 56 Marine humic acids Mean 56.3* 5.8* 3.8 3.1" 31,7 0.45 1.23 Median 56.4 5.5 3.7 2.6 32.1 0.43 1.20 Mde 55.9-57.6 (21) 5.1-5.4 (14) 3.0-3.8 (26) 2.2-2.5 (9) 31.8-35.6 (34) 0.40-0.45 (23) 1.05-1.11 (12) Range 37.52-75.76 3.76-11.68 0.9%10.54 1.2-8.3 7.93-56.6 0.08-1.20 0.6%1.85 SD 6.6 (11.7) 1.4 (24.1) 1.5 (39.5) 1.4 (45.2) 7.8 (24.6) 0.18 (40.0) 0.23 (18.7) X2-statistic 14.0 (14.1) 14.4 (14.1) 6.5 (14.1) 3.3 (9.5) 23.7 (14.1) 33.9 (14.1) 7.8 (14.1) N. f samples 95 95 95 66 95 95 95 Peat humic acids Mean 57,1 5.0 2.8 0.4* 35.2 0.47 1.04 Median 57.4 5.0 2.9 0.3 34.5 0.46 1.06 Mde 57.7-58.4 (5) 5.0-5.2 (4) 3.3-3.5 (5) 1.9-2.1 (5) 35.8-36.5 (5) 0.47-0.48 (7) 1.04-1.13 (6) Range 50.53-62.75 3.6-6.57 0.60-3.9 0,1-0.7 30.68-43.20 0.37-0.64 0.73-1.35 SD 2.5 (4.4) 0.8 (16.0) 1.0 (35.7) 0.2 (50.0) 2.7 (7.7) 0.06 (12.8) 0.17 (16.3) X 2-statistic....... N. f samples 23 23 21 12 23 23 23 An asterisk (*) indicates that this parameter in this humic acid exhibited r mst-clsely apprximated a lg-nrmal distributin; all ther values exhibited a nrmal distributin. Values in parentheses fllwing the mde are the number f samples in that interval, thse fllwing the standard deviatin are relative standard deviatins (RSD). x2-statistics are prvided fr the mst nrmal frm f the data as indicated abve and X 2 fr that sample size fllws in parentheses. e~ I~'. E 8 m
642 JAbl~ A. RICE and PATRICK MACCARTHY prbability level. This culd be the result f real similarities in the average elemental cmpsitins f humic acid and humin (n an ash-free basis) r it culd be due t the small number f analyses frm which the humin data set was cnstructed. A small data set tends t diminish the ability f statistical devices, such as the t-test, t discern subtle differences. Of curse, humic acid and humin differ cnsiderably in ther characteristics such as ash cntent, slubility, and s n. Humic acids segregated by surce (I) General trends in elemental cmpsitin fr humic acids. Table 3 cntains the statistical data fr humic acids where the samples are segregated accrding t surce: sil, freshwater, marine, and peat. Cal-derived humic acids, which were included in Table 1, are nt included in this classificatin as there were t few data t make the presentatin f average elemental cmpsitins meaningful fr this categry r t perfrm tests f statistical significance. The distributin f the data, n the basis f X2-stat - istics, mre clsely apprximates nrmality when the humic acids are segregated by surce (Table 3) than when they are cnsidered in tt (Table 1). The mst striking bservatin n cmparing the data in Table 3 with thse in Table 1 is that when the humic acids are segregated by surce, the range f carbn values narrws dramatically and the standard deviatins fr carbn decrease remarkably except fr the marine humic acid, where nly a small decrease in the range ccurs and an increase in the relative standard deviatin is bserved. Fr example, the standard deviatins fr carbn cntents decrease by abut 25, 40 and 50% fr sil, freshwater and peat humic acids, respectively, cmpared t the standard deviatin f humic acids cnsidered in tt. This bservatin, in cnjunctin with the nticeably different averages fr humic acids frm different surces (Table 3) suggests a "preferred" cmpsitin fr the humic acids frm each envirnment. The fact that the standard deviatins d nt cntract in the case f the marine humic acids may result frm the variety f envirnments (near-shre, deepsea, etc.) in this study which were classified by the riginal authrs as being marine, sme f which have the likelihd f several surces f rganic matter input. Fr example, a near-shre marine envirnment may receive rganic matter frm terrestrial as well as algal surces due t its prximity t land masses. Table 4. t-statistics, and their significance, used t cmpare humic acids frm sil, freshwatvr and marine surces frm Table 2 t-statistic t~ C H N O O/C H/C Sil-freshwater 3.39 8.98* 0.19 6.07* 8.43* 8.38* 2.68 Sil-marine 3.40 0.79 7.37* 0.71 4.83* 3.70* 6.01" Freshwatermarine 3.29 6.56* 8.80* 5.01" 10.82" 10.15" 4.14" tc Represents the critical t-value at P = 0.0005 fr each cmparisn and an asterisk (*) indicates a t-value significant at P < 0.0005. There is a similar prnunced cntractin (except fr the marine samples) f the standard deviatin fr xygen cntents when the humic acids are separated by surce. This is als generally true fr the standard deviatins f S, N and H (with sme exceptins) thugh the effect is nt s prnunced. The larger H/C rati (Table 3), cnsistent with a mre aliphatic nature, fr marine humic acid relative t sil humic acid is in agreement with previus reprts based n cnsiderably smaller data sets (Ishiwatari, 1969; 1970; Rashid and King, 1970; Stuermer and Payne, 1976; Steelink, 1985). The average elemental cmpsitin f sil humic acid in Table 3 falls within the range f values reprted fr an "average" sil humic acid by Schnitzer (1977) based n a substantially smaller data set. (2) Statistically significant differences in humic acids. Table 4 presents the statistical significance f differences in the means f the elemental cmpnents f humic acid islated frm different surces. Peat humic acid was nt invlved in this cmparisn because f the cmparatively small number f data in this set. Table 4 shws that humic acid btained frm freshwater surces pssesses an elemental cmpsitin distinct frm sil and marine humic acids. Marine humic acid is statistically different frm sil humic acid with respect t hydrgen and xygen cntents and the H/C rati. The O/C rati suggests that real differences exist between freshwater, sil and marine humic acids. A statistical distinctin may be made between marine humic acid and freshwater and sil humic acids n the basis f the H/C rati. A similar distinctin cannt be made between sil and freshwater humic acids at the chsen prbability level, althugh they are statistically different at a lwer level, P < 0.005. This lwer prbability level still allws nly ne chance in 200 that the means culd be bserved in the same ppulatin, and suggests that the lack f statistical distinctin between the freshwater and sil humic acids at P < 0.0005 may be due t the small number f analyses f the freshwater humic acid relative t the sil humic acid. Thus, it is prbably reasnable t infer that the aliphaticity f humic acids frm different surces decreases in the rder marine > freshwater > sil. This wuld be cnsistent with the nature f the rganic input int sil and marine envirnments. Marine rganic detritus is believed t be predminantly algal in rigin and f a highly aliphatic nature (Nissenbaum and Kaplan, 1972) while sil rganic input is believed t be mre armatic in nature due t the ubiquity f lignin in terrestrial plants (Flaig, 1972). A freshwater sample culd have majr inputs f either algal r sil rganic materials. In summary, the data in Table 4 indicate that there are statistically significant differences between the elemental cntents f humic acids frm sil, freshwater and marine surces.
Table 5. Mean, median, mde, range, standard deviatin (SD) and X2-statistics fr elemental cmpsitins f fulvic acids frm different surces expressed as weight percent, O/C and H/C ratis are atmic percent. All values are n an ash-free basis C H N S O O/C H/C Sil fumc acids Mean 45.3 5.0 2.6 1.3 46.2 0.78 1.35 Median 44.2 5.0 2.8 0.9 47.1 0.79 1.33 Mde 39.9-41.3(17) 4.8-5.0(18) 0.7-1.0(18) 0.2~).4 (13) 43.6-44.9(19) 0.824).86(17) 1.00-1.09(16) Range 35.1-75.4 3.2-7.00 0.45-5.87 0.1-3.6 16.9-55.88 0.17-1.19 0.77-2.13 SD 5.4 (11.9) 1.0 (20.0) 1.3 (50.0) 1.1 (84.6) 5.2 (11.3) 0.16 (20.5) 0.34 (25.2) Z2-statistic 36.7 (18.3) 13.3 (I 8.3) 33.5 (18.3) -- 20.5 (18.3) 18.8 (18.3) 4.8 (18.3) N. f samples 127 127 127 45 127 127 127 Freshwater fulvic acids Mean 46.7 4.2 2.3 * 1.2 45.9* 0.75 I. 10* Median 46.2 4.3 1.8 1.0 46.8 0.76 1.10 Mde 44.5-45.6(11) 4.2~.4 (13) 1.8-2.3(16) -- 48.3--49.7(10) 0.78-0.81 (8) 1.09-1.12(11) Range 39.2-56.33 0.43-5.9 0.47-8.16 0.1 ~3.05 34.66-55.8 0.49-1.07 0.81-1.53 SD 4.3 (9.2) 0.7 (16.7) 2.1 (91.3) 0.9 (75.0) 5.1 (11.1) 0.14 (18.7) 0.13 (11.8) 12-statistic 8.9 (9.5) 12.9 (9.5) 20.5 (9.5) -- 8.9 (9.5) 4.3 (9.5) 3.2 (9.5) N. f samples 63 63 63 14 63 63 63 Marine fulvic acids Mean 45.0 5.9 4.1 2.1 45.1 0.77 1.56 Median 45.4 6. l 4.5 -- 44.5 0.73 1.62 Mde....... Range 38.4-50.00 4.34i.80 1.0~.83 -- 36.9-54.5 0.55-1.07 1.31-1.73 SD 4.0 (8.9) 0.9 (15.3) 2.3 (56.1) -- 6.0 (13.3) 0.17 (22.1) 0.13 (8.3) X2.stafistic....... N. f samples 12 12 12 1 12 12 12 Peat fulvic acids Mean 54.2 5.3 2.0 0.8 37.8 0.53 1.20 Median 54.1 4.9 2.2 0.6 38.8 0.54 1.04 Mde....... Range 46.9-60.8 4.2-7.2 1.2-2.6 0.2-1.9 31.1-44.3 0.38-0.71 0.85-1.84 SD 4.3 (7.9) 1.1 (20.8) 0.5 (25.0) 0.6 (75.0) 3.7 (9.8) 0.094 (17.0) 0.33 (27.5) Z 2 statistic....... N. f samples 12 12 12 11 12 12 12 An asterisk (*) indicates that this parameter in this fulvic acid exhibited r mst-clsely apprximated a lg-nrmal distributin; all ther values exhibited a nrmal distributin. Values in parentheses fllwing the mde are the number f samples in that interval, thse fllwing the standard deviatin are relative standard deviatins (RSD). x2-statisties are prvided fr the mst nrmal frm f the data as indicated abve and X 2 fr that sample size fllws in parentheses. g 8 ~ ~r B" ga m
644 J~s A. RIcE and PATmCK MACCAItTm Fulvic acids segregated by surce (1) General trends in elemental cmpsitin fr fulvic acids. Table 5 cntains the statistical data fr fulvic acids segregated by surce. Gruping fulvic acids by surce causes a significant narrwing f the range and standard deviatin f carbn values, except in the case f the sil samples where the range is unaltered and the relative standard deviatin is virtually unchanged. The ranges and standard deviatins fr mst ther parameters als decrease when the fulvic acids are segregated by surce althugh there are exceptins t this generalizatin. Segregatin f the fulvic acid samples by surce (Table 5) results in the data mre clsely apprximating a nrmal distributin than when thse data are evaluated in tt (Table 1). The mst nrmal frm f the data in Table 5, in cntrast t that in Table 1, exhibits Z2-statistics very near r belw the critical values, cnfirming that these data exhibit r very clsely apprximate a nrmal distributin. Cmpared t sil, freshwater and marine fulvic acids, all f which exhibit similar carbn and xygen cntents, peat fulvic acid displays markedly different values fr C (higher), O (lwer) and cnsequently, a lwer O/C rati. Hydrgen and H/C values fr peat fulvic acid fall within the range f values exhibited by ther samples. Hwever, cautin must be exercised in attempting t draw cnclusins frm these bservatins in that nly twelve samples f peat fulvic acid are invlved in this study and eleven f these were cllected by a single grup frm different areas f ne lcale (Zelazny and Carlisle, 1974). This may intrduce sme bias int the estimate f this sample ppulatin mean, and cnsequently peat fulvic acid will nt be discussed any further in this paper. Cmparing the sil, freshwater and marine "average" fulvic acids in Table 5 reveals that the C and O cntents and the O/C ratis are quite similar. Thugh nt as prnunced as when humic acids are segregated by surce, the H, N and S cntents f the fulvic acids d shw different averages, when gruped by surce, cmpared t the average f all fulvic acids cmbined. On the basis f the H/C rati the "average" sil fulvic acid wuld appear t be mre aliphatic than freshwater fulvic acid. (2) Statistically significant differences in fulvic acids. Table 6 reveals that the nly statistically significant difference between fulvic acids frm sil and freshwater envirnments is in the hydrgen cntent and in the H/C rati. Because f the small number f marine and peat fulvic acids fr which data were available, they were excluded frm this type f cmparisn. By Table 6. t-statistics, and their significance, used t cmpare fulvic acids frm sil and freshwater surces frm Table 3 t -Statistic t c c H N O O/C H/C Sil-freshwater 3.29 2.56 6.88* 2.23 0.30 2.04 7.18" t, Represents the critical t-value at P = 0.0005 and an asterisk (*) indicates a t-value significant at P < 0.0005. Table 7. t-statistics, and their significance, used t cmpare humic acids (HA) and fulvic acids (FA) frm similar (sil and freshwater) surces t -Statistic t C H N O O/C H/C Sil HA-FA 3.29 16.75" 2.19 8.24* 17.85" 17.41' 8.36* Freshwater HA-FA 3.38 6.35* 4.35* 2.47 6.04* 6.42" 0.72 t c Represents the critical t-value at P = 0.0005 and an asterisk (*) indicates a t-value significant at P < 0.0005. inference the~, sil fulvic acid is mre aliphatic than freshwater fulvic acid due t its larger H/C rati. This result is interesting because it has ften been stated that sil humic materials are generally mre armatic than humic substances frm ther envirnments. Fr example, Visser (1983) bserved that humic materials frm freshwater surces generally exhibited H/C ratis greater than that f sil humic substances. This apparent difference is usually attributed t the ubiquity f lignin, which sme believe t be an imprtant humic precursr, in terrestrial plants. The H/C rati fr freshwater humic acid is greater than that fr sil humic acid (Table 1) but the difference is nt statistically significant at P < 0.0005 (Table 5). Intercmparisn f humic and fulvic acids frm similar envirnmen ts (1) General trends in elemental cmpsitin. When segregated by surce, fulvic acids (Table 5) cnsistently exhibit a lwer C cntent, higher O cntent and a higher O/C rati than d humic acids (Table 3) frm the same type f envirnment. N clear distinctins between humic acid and fulvic acid frm similar envirnments are evident when trends in the H, N 2.5 2.0 1.5 "1- i ~. 0..5 HA %, fyiti FA ".%',% HMN I 0.5 1.0 1.5 Atmic O/C rti Fig. 2, van Krevelen diagram based n humic acid (410 samples), fulvic acid (214 samples) and humin (26 samples).
Statistical evaluatin f the elemental cmpsitin f humic substances 645 and S cntents are examined. The fulvic acids, with ne exceptin, exhibit a greater H/C rati than the respective humic acids. (2) Statistically significant differences between humic and fulvic acids frm similar envirnments. Visser (1983) cncluded that the majrity f fulvic and humic acid samples frm freshwater envirnments exhibit similar H/C ratis. The results f the t-test in Table 7 supprt this bservatin by indicating that the differences in the means f the H/C ratis in freshwater fulvic acid and humic acid are nt significant. Hwever, freshwater humic acid and fulvic acid differ significantly frm each ther with respect t the ther parameters in Table 7, except fr the nitrgen cntent. Table 7 als crrbrates statements in the literature (Knnva, 1966; Schnitzer and Skinner, 1974) which pint ut that sil fulvic acid (Table 5) is mre aliphatic than sil humic acid (Table 3). Prestn and Ripmecster (1982) reprt a "predminantly aliphatic" nature fr sil fulvic acid based n ~3C-NMR spectra. Sil fulvic acid als exhibits a statistically different weight-percent, with respect t sil humic acid, fr every element except hydrgen (Table 7) emphasizing that these tw sil humic substances differ in mre ways than just their degree f aliphaticity. Because f the small ttal number f humin samples in the data set (Table 1), humin was nt segregated by surce. van Krevelen diagrams Figure 2 shws a van Krevelen diagram based n the 650 samples in Table 1, and illustrates the cmp- sitinal fields ccupied by humic acid, fulvic acid and humin. In this paper van Krevelen diagrams are used, as elsewhere (Kuwatsuka et al., 1978; Visser, 1983), nly t summarize cmpsitinal differences between humic materials. Humic and fulvic acids ccupy brad cmpsitinal fields with a cnsiderable regin f verlap. The humin samples ccupy a cmpsitinal field that lies almst cmpletely within that f humic acid. The "thumb" extending frm the humic acid field int the upper left-hand prtin f Fig. 2 represents 11 marine humic acids extracted frm Ecene chalks and cherts in the investigatin f Bein and Sandier (1983). These samples exhibit H/C ratis which are nt markedly different frm thers in the humic acid data set; the O/C ratis f these samples, hwever, represent the lwer extreme in the range f O/C ratis listed in Table 1 shwing values f abut 0.20 r less. Even thugh these 11 humic acids were extracted by a rather cnventinal sdium hydrxide-sdium pyrphsphate prcedure (Bein and Sandier, 1983), and cnsequently cnfrm t the peratinal definitin f humic acid, sme bvius differences exist between them and the bulk f the humic acids represented in this diagram. The van Krevelen diagram presented in Fig. 2 differs frm that f Kuwatsuka et al. (1978) in three ways: (1) this diagram includes data fr humin samples, (2) apprximately ten times as many humic acid samples and thirty times as many fulvic acid samples are utilized in the present diagram and (3) Kuwatsuka et al. (1978) described the distributin f 2.5 2.5 2.0 2.0... 222 "~,~:~i :*: ~:... ~s*; t.s*. :'._ :. ~ ~... 1.0,...,.~:.......:::!:;::::. E 1.5 -r E 0 1.0.5 ~ ii~ E~ M 0.5 fft~lm ~s rrrnp I I 0.5 1.0 1.5 Atmic O/C rti Fig. 3. van Krevelen diagram based n sil humic acids (215 samples), freshwater humic acids (56 samples), marine sediment humic acids (95 samples) and peat humic acids (23 samples). I 0.5 1.0 Atmic O/C rti ~s FTmp Fig. 4. van Krevelen diagram based n sil fulvic acids (127 samples), freshwater fulvic acids (63 samples), marine sediment fulvic acids (12 samples) and peat fulvic acids (12 samples). I 1.5
646 JAMES A. RICE and PATRICK MACCARTHY the humic acids n their van Krevelen diagram as being '"J"-shaped', but such a descriptin is nt applicable t the larger data set emplyed in the cnstructin f Fig. 2 even thugh the data f Kuwatsuka et al. are included in this figure. In general, hwever, the regins ccupied by humic and fulvic acids are similar t thse f Kuwatsuka et al. (1978). The cnsiderably larger data base used in the present study enhances the reliability f cnclusins based n these diagrams. Tisst and WeRe (1978) have described humic substances as being similar t Type III kergens. In terms f Fig. 2 and the data in Table 1 (mderate H/C ratis, high O/C ratis), this appears reasnable; the average humic material exhibits atmic ratis typical f Type III kergens. Figures 3 and 4, which are van Krevden diagrams fr humic acids and fulvic acids, respectively, segregated by surce, emphasize the difference between humic materials frm different envirnments that were described in earlier sectins. Figure 3 shws that freshwater, peat and marine humic acids exhibit cnsiderable verlap in their cmpsitins. The sil humic acid cmpsitinal field verlaps that f the ther three. The cmpsitinal fields ccupied by freshwater, marine and peat fulvic acids (Fig. 4) appear t be mutually mre distinct than their humic acid cunterparts (Fig. 3). Such apparent distinctins may nt be evident if the marine and peat fulvic acid data sets were larger. As with humic acids segregated by surce (Fig. 3), sil fulvic ac;d extensively verlaps the cmpsitinal fields f the ther three fulvic acids in Fig. 4. Average elemental cmpsitins f artificial and sewage sludge "humic " materials Finally, in additin t the natural humic substances already discussed in this paper, the elemental data fr a very limited number f artificial and sewage-sludge derived "humic materials" were als briefly examined fr cmparative purpses. The mean values fr the dcmental cntents and atmic ratis f 17 micrbially-prduced "humic acids", 3 chemically synthesized "humic acids", and 3 sewagesludge "fulvic acids" are given in Table 8. The micrbially-prduced "humic acids" have mean values similar t thse fr unsegregated humic acids (Table 1). Fr the chemically synthesized "humic acids" the O/C rati is similar t, and the H/C rati is less than that f unsegregated humic acids. Only ne f these three "humic acids" cntained nitrgen (at a very high level). Chemically synthesized humic Table 8. Mean elemental cntents (weight pcrcant) and atmic ratis fr i 7 micrbially prduced "humic acids", 3 chemically synthesized "humic acids" and 3 sewage sludge "fulvic acids". All values an: n an ash-free basis Sample C H N S O O/C H/C Micrbial 52.6 5.1 4.1 -- 38.1 0.54 1.16 Chemical 58.1 3.2 3.5 -- 37.4 0.48 0.66 Sewage-sludge 40.82 6.57 2.83 8.15 42.27 0.78 1.93 acids d nt cntain nitrgen unless it is present in ne f the reagents. The O/C rati fr the sewage sludge "fulvic acid" is similar t that f the unsegregated fulvic acids (Table 1) while the H/C value is cnsiderably greater. The sewage sludge "fulvic acids" have a very high sulfur cntent pssibly due t the presence f sulfnated detergents; all three f these "fulvic acids" cme frm the same study (Spsit et al., 1976). Differences between cmmercial humic acids and sil and water humic substances have als been reprted (Malclm and MacCarthy, 1986; MacCarthy and Malclm, 1989). SUMMARY The statistical analysis presented in this paper quantitatively identifies similarities and differences between varius humic substances, and in additin t establishing new trends in the prperties f these materials, it prvides a statistical basis fr sme previusly-reprted trends. The results f this study are summarized as fllws: (1) Humic acid, fulvic acid and humin exhibit surprisingly small standard deviatins in their elemental cntents, particularly fr carbn, even fr unsegregated samples. This may suggest that sme preferred cmpsitin, r a relatively narrw range f cmpsitins, exists fr humic substances in nature. (2) The t-test indicates that the C, N, and O cntents and the O/C and H/C ratis exhibited by fulvic acid are significantly different frm thse f humic acid at P < 0.0005. The carbn and xygen cntents and the O/C rati f fulvic acid are statistically different frm thse f humin. N statistically significant differences in the elemental cmpsitin exist, n the basis f these data sets, between humin and humic acid. (3) When segregated by surce, humic acids frm different envirnments exhibit different means in the elemental parameters; cnsiderable decreases in standard deviatins cmpared t humic acids in tt are evident. (4) Freshwater humic acid exhibits statistically significant differences in its C, N and O cntents and O/C rati cmpared t marine and sil humic acids. Marine humic acids display statistically significant differences in their H and O cntents and in the O/C and H/C ratis cmpared t sil humic acids. In additin, tw f the three humic acid surce intercmparisns (marine-sil and marine-freshwater) exhibit statistically different H/C ratis. By inference, the degree f aliphaticity f humic acids decreases in the rder marine > freshwater > sil. (5) Variatins in means f the fulvic acids, when segregated by surce, are nt as prnunced as fr humic acids, but in mst cases there is still a cnsiderable decrease in the standard deviatins cmpared t unsegregated fulvic acids.
Statistical evaluatin f the elemental cmpsitin f humic substances 647 The dimunitin f the standard deviatin which ccurs when the samples are segregated by surce, cupled with the statistically significant differences in the mean indicates, perhaps, that there is an ptimum cmpsitinal range t be expected fr each humic material in a particular envirnment. (6) Statistically distinct H cntents and H/C ratis exist in fulvic acids frm sil cmpared t freshwater surce envirnments. Because the H/C rati exhibited by the sil fulvic acids is larger, it is assumed t be mre aliphatic than freshwater fulvic acid. (7) Sil fulvic acid exhibits statistically significant differences in the C, N and O cntents and O/C and H/C ratis when cmpared t thse f sil humic acid. (8) A van Krevelen diagram based n 650 samples f humic acids, fulvic acids and humins shws that humic and fulvic acids ccupy distinct regins n the diagram with an area f verlap. The humin cmpsitinal field exists almst cmpletely within the humic acid field. (9) A van Krevelen diagram based n humic acids segregated by surce shws cnsiderable verlap between the cmpsitinal fields f sil, freshwater, marine and peat humic acids. A van Krevelen diagram based n fulvic acids segregated by surce shws that sil fulvic acid verlaps the cmpsitinal fields f freshwater, marine and peat fulvic acids. Freshwater, marine and peat fulvic acids appear t exhibit distinct cmpsitinal fields n the diagram. (10) While chemically synthesized "humic acids" and sewage sludge "fulvic acids" cnfrm t the peratinal definitins f these materials their minr elemental cmpnents indicate differences between them and naturally ccurring humic substances. Sewage sludge "fulvic acids" and chemically synthesized "humic acids" exhibit sme prnunced differences in their S and N cntents, respectively, cmpared t "natural" humic materials. Micrbially prduced humic acids are generally similar t the humic acid data set in tt. REFERENCES Bartlett M. S. (1947) The use f transfrmatins. Bim. Bull. 3, 39-51. Bein A. and Sandier A. (1983) Early diagenetic xidatin and maturatin trends in rganic matter extracted frm Ecene chalks and cherts. Chem. Gel. 38, 213-224. Crw E. L, Davis F. A. and Maxfield M. W. (1960) Statistics Manual. Dver, New Yrk. Flaig W. (1972) Bichemical factrs in cal frmatin. In Advances in Organic Gechemistry 1971 (Edited by vn Gaertner H. R. and Wehner H.), pp. 197-232. Pergamn Press, Oxfrd. Hue A. Y. and Durand B. M. (1977) Occurrence and significance f humic acids in ancient sediments. Fuel 56, 73-80. Huffman Jr E. W. D. and Stuber H. A. (1985) Analytical methdlgy fr elemental analysis f humic substances. In Humic Substances in Sil, Sediment, and Water: Gechemistry, Islatin, and Characterizatin (Edited by Aiken G. R., McKnight D. M., Wershaw R. L. and MacCarthy P.), pp. 433~,55. Wiley, New Yrk. Ishiwatari R. (1969) Fractinatin and characterizatin f humic acid frm a lake sediment. Gechem. J. 2, 175-184. Ishiwatari R. (1970) Structural characteristics f humic substances in recent lake sediments. In Advances in Organic Gechemistry (Edited by Hbsn G. D. and Speers G. C.), pp. 283-311. Pergamn Press, Oxfrd. Knnva M. M. (1966) Sil Organic Matter (Translat. Nwakwski T. Z. and Newman A. C. D.). Pergamn Press, Oxfrd. Klusman R. W. (1981) PHIST, a FORTRAN prgram t calculate a summary statistics. Unpublished research, Clrad Schl Mines, Glden, Cl. Klusman R. W., Ringrse C. D., Candit R. J., Zuccar B. and Dean W. E. (1980) Sampling designs fr gechemical baseline studies in the Clrad Oil Shale regin: a manual fr practical applicatin. DOE/EV/10298-2, U.S. Department f Energy, Washingtn, D.C. (available, NTIS). van Krevelen D. W. (1961) Graphical-statistical methd fr the study f structure and reactin prcesses f cal. Fuel 29, 269-284. Kuwatsuka S., Tsutsuki K. and Kumada K. (1978) Chemical studies n sil humic acid I. Elementary cmpsitin f humic substances. Sil Sci. Plant Nutr. 24, 337-347. MacCarthy P. and Malclm R. L. (1989) The nature f cmmercial humic acids. In Aquatic Hurnic Substances: Influence n Fate and Treatment f Pllutants (Edited by Suffet I. H. and MacCarthy P.), pp. 55-63. American Chemical Sciety, Washingtn, D.C. Malclm R. L. and MacCarthy P. (1986) Limitatins in the use f cmmercial humic acids in sil and water research. Envirn. Sci. Technl. 20, 904-911. Nissenbaum A. and Kaplan I. R. (1972) Chemical and istpic evidence fr the in situ rigin f marine humic substances. Limnl. Oceangr. 17, 570-582. Perdue E. M. (1984) Analytical cnstraints n the structural features f humic substances. Gechim. Csmchim. Acta 48, 1435-1443. Prestn C. M. and Ripmeester J. A. (1982) Applicatin f slutin and slid-state ~C NMR t fur rganic sils, their humic acids, fulvic acids, humins and hydrlysis residues. Can. J. Spectrsc. 27, 99-105. Rashid M. A. and King L. H. (1970) Majr xygen-cntaining functinal grups present in humic and fulvic acid fractins islated frm cntrasting marine envirnments. Gechim. Csmchim. Acta 34, 193-202. Reuter J. H. and Perdue E. M. (1984) A chemical structural mdel f early diagenesis f sedimentary humus/prtkergens. Mitt. Gel. Paldntl. lnst. Univ. Hamburg. 56, 249-262. Satterthwaite F. E. (1946) An apprximate distributin f estimates f variance cmpnents. Bita. Bull. 2, 110-114. Schnitzer M. (1977) Recent findings n the characterizatin f humic substances extracted frm sils frm widely differing climatic znes. In Sil Organic Matter Studies, Vl. 2, pp. 117-132. Internatinal Atmic Energy Agency, Vienna. Schnitzer M. and Skinner S. I. M. (1974) Lw temperature xidatin f humic substances. Can. J. Chem. 52, 1072-1080. Spsit G., Hltzclaw K. M. and Baham J. (1976) Analytical prperties f the sluble, metal-cmplexing fractins in sludge-sil mixtures---ii. Cmparative structural chemistry f fulvic acid. Sil Sci. Sc. Am. J. 40, 691-697. Steelink C. (1985) Implicatins f elemental characteristics f humic substances. In Humic Substances in Sil, Sediment, and Water: Gechemistry, Islatin, and Characterizatin (Edited by Aiken G. R., McKnight D. M., Wershaw R. L. and MacCarthy P.), pp. 457--475. Wiley, New Yrk.