I t . ; c , R .~u td~ kol I0. p p gt)5 t,., 811 Pergamon Press 19'~6 Prtnted m Great B m a i n

THE RELATION OF FLUORESCENCE TO DISSOLVED ORGANIC CARBON IN SURFACE WATERS

P. L. SMART

Department of Geography. Universit3 of Bristol. Bristol. England

B. L. FINLAYSON

School of Geography. Oxford University. Oxlord. England

and

W. D. RYLANDS and C. M. BALL

Department of Geography. University of Bristol. Bristol. England

tReceived 12 February 19761

Abstrnct--A rapid method for the estimation of organic carbon in natural waters which could be adapted to continuous measurement is needed to improve water quality monitoring. The fluorescent properties of natural organic molecules offer such a method. The fluorescence of natural waters from a variety of surface sources, using an excitation wavelength of 365 nm and emission wavelengths in the range 400-600nm have been correlated with total organic carbon levels. The results are highly significant statistically and show that fluorescence can be used as a predictor of TOC over a wide range of concentrations.

I . I N T R O D U C T I O N

The monitoring of organic substances in natural waters has long been of importance to organizations concerned with water supply and waste disposal. There are maximum acceptable levels of such sub- stances for domestic water supply (U.S. Public Health Service. 1961)and for many industrial uses (Abrams, 1969). The use of rivers for the discharge of sewage effluents, industrial and agricultural wastes and drainage from roads has led to serious biological con- sequences due to the effect of organic substances on dissolved oxygen levels, and to the introduction of toxic compounds. Goerlitz and Brown (1971 p. 1) write that "the value of even the most primitive measurements of organic substances in water is apparent, considering the general awareness of man's effect on the environment and the implementation of programmes to stem the tide of water pollution, lake eutrophication and similar problems". Although natural organic load may vary relatively slowly, waste water discharges often produce rapid fluctuations in total organic carbon {TOC). These will often not be satisfactorily defined where periodic samples alone are collected, thus a system permitting continuous field monitoring would be of considerable utility.

Dissolved organic matter is also of major impor- tance in the biological and earth sciences. It is a basic component in the carbon and energy budgets of stream and lake ecosystems (Fisher and Likens, 1972). Similarly it is an important chemical parameter, rele- vant in studies of geochemistry, (Ling Ong et al., 19701. water quality (Baker et al., 1974) and in denu- dation studies (Viro. 19531. In many of these fields

extensive sampling programmes combined with con- tinuous monitoring would be desirable (Manny and Wetzel, 1973; Walling, 1975). Furthermore, operation under field conditions might be necessary where laboratory facilities are not locally available.

Standard methods for the determination of levels of organic matter in water are the Chemical Oxygen Demand (COD) and Biochemical Oxygen Demand (BOD) tests (American Public Health Association, 1965; Society for Analytical Chemistry, 1958). Both methods are slow and tedious, the results are not easily reproducible and are difficult to interpret. The use of these tests severely limits the number of samples which can be analysed. However, the test for TOC (Van Hall et al.. 1963; Baker et al.. 1974) is quick, precise, and the results are easily interpreted since all the organic carbon present in the sample is oxidized. An analysis rate of about 10 samples per hour is possible using this method. Unfortunately the equipment is expensive, and may not be justified un- less very heavy regular use is possible.

A number of workers have experimented with the use of u.v. absorbance to monitor the TOC of water (Dobbs et al., 19721. This method does not give the accuracy obtainable using TOC analysis, but has the advantage of using readily available instrumentation and being relatively simple. More recently Mattson et al. (1974} have presented results from a prototype double-beam u.v./visible photometer for continuous field use. Mrkva (1975) has investigated the use of absorbance at 254 nm for monitoring organic matter in various waters with calibration against permanga- nate oxygen demand and dichromate oxygen demand which are estimates of COD.

805

800 P.L. SMART. B. L. FINLAYSON. W. D. RYLANDS and C. M. BALL

This paper reports the preliminary results of exper- iments on the use of fluorescence for monitoring the TOC of natural waters. A comparison of fluorescence and u.v. absorbance measurements is also presented.

2. PREVIOUS ~ O R K

Dienert (1910) observed the fluorescence of natural waters, but little use has been made of this property except for the separation of components in chromato- graphy using ultra-violet light [Sowden and Deuel. 1961). A number of fluorescent spectra have been reported in the literature with excitation peaks from 360 nm (Ghassemi and Christman, 1968), to 470 nm (L'evesque, 1972) and in the range 346-427 nm (Hall and Lee, 1974). Fluorescent emission spectra are generally of considerable band width with peaks reported at 450 nm (Ghassemi and Christman, 1968: Levesque, 1972), 520nm (Seal et al.. 1964; Black and Christman, 1963) 426-510nm (Hall and Lee, 1974), and 405-410 nm (Wilander et al., 1974). Ghassemi and Christman {1968) reported that the spectra were iden- tical for natural waters and those obtained after mol- ecular size separation on Sephadex gel. though the high molecular weight components appeared to be less fluorescent. Christman (1970) has observed that the fluorescent spectra of many natural waters are remarkably similar, however it is possible that some differences may be caused by both origin and sample preparation.

Black and Christman (1963) reported that the organic carbon to fluorescence ratio was not constant for the diverse waters they examined. Ghassemi and Christman (1968) support this finding, but L'evesque. (1971) has presented a satisfactory calibration curve of fluorescence against fulvic acid concentration. It was found that internal quenching or "inner filter" effects caused non-linearity above 40 mg I-~ giving a quadratic-like response.

Natural waters contain a variable ratio of "'humic" and "'fulvic" acids which may effect their fluorescence. Prakash (in Discussion of Christman, 1970) supports Packham (1964) in stating that fulvic acid contains smaller molecules than humic acid. He suggests that this may explain why differences in fluorescence per unit carbon are obtained for these two components. If such differences are significant it may be necessary to resort to specific calibration curves such as those of Christman and Minear 11967) and Wilander et al. t1974) for lignin sulphonates. Both groups found that a good calibration could be obtained between fluores- cence and lignin sulphonate concentration. Baum- gartner et al. (1971) used fluorescence monitoring techniques in the field to study dispersion of pulp mill waste in receiving waters. Sewage may also pos- sess particular characteristics at a specific site, as noted by Jones {1969L who showed that a constant ratio between TOC and BOD, or TOC and COD was only present through time at individual outfalls.

It is apparent from the literature that a relationship between organic matter and fluorescence in water may exist. However, it is clear that there may be con- siderable variation in fluorescent spectra and that some organic matter may not fluoresce at all, as shown by chromatographic separation (Sowden and Deuel, 1961: L'evesque. 197l). It will thus be necess- ary to examine these problems carefully if any useful technique is to be developed. Furthermore. tempera- ture (Christman and Minear, 1967), pH (Black and Christman, 1963) and the presence of metals, even in low concentrations may affect the fluorescence of natural waters (Ghassemi and Christman, 1968). The optical properties of coloured waters exhibit poor sta- bility ill'in and Orlov, 1973J. Lamar and Goerlitz (1966) report that darkening, lading, and formation of precipitates may occur in water samples after col- lection, so that sample preparation and preservation must also be considered.

3. SAMPLE COLLECTION AND PREPARATION

A variety of waters were collected from the Bristol region in 500 ml glass bottles. Plastic bottles may con- tribute organic material to the water sample and should not be used. The samples were filtered through 0.4 ~Lm membrane filters immediately after collection. They were kept in the dark and below 5°C during further storage.

Three groups of samples were collected, all from the Bristol region, England.

Group 1

Samples were collected in a single sampling oper- ation from a variety of sources including karst spring waters, a variety of polluted and unpolluted river waters, soil waters, agricultural wastes and a sample of sea water. The samples within this group had pH values between 6.7 and 8.4, with an average of 7.8. No attempt was made to buffer them to a standard pH value.

Group 2

Samples were collected from a motorway drain on the M4 Bristol-London Motorway. The samples were all analysed without buffering.

Group 3

Samples were collected over several months from the East Twin basin (Mendip Hills. Somerset). a small natural upland catchment developed on Old Red Sandstone sandstones and shales. The upper part of the basin has peaty soils with a heather community, while the lower slopes have brown earths with bracken and moorland grasses. Samples were col- lected from soil waters in the upper and lower basin and from the stream. Because of the wide range of pH values encountered 13.2-7.8) the samples were buf- fered to pH 7.0 using B.D.H. buffer solutions prior

Dissoh'ed organic carbon in surface waters S07

to analysis. This follows the procedure of Ghassemi and Christman [ 1968t.

4. INSTRUMENTAL TECHNIQUES

Excitation and emission spectra were determined using an Aminco-Bowman Ratio Spectrofluorometer fitted with an xenon lamp. From this data filters were selected for a Turner 111 filter fluorometer which was to be used for routine determinations. The primary filter was a Coming Cs %83 and the secondary a Kodak Wratten No, 64. The fluorometer was equipped with a low pressure mercury lamp (General Electric Co G4T4.11 having significant emission only at the mercury lines, and a temperature controlled door. All samples were brought to 2ZC in a water bath before analysis. The pH of each sample was measured on a Pye Unicam 292 pH meter.

Absorbance was measured on a Unicam SP800 u.v./visible spectrophotometer using the method of Dobbs et al. (19721. A Phase Separations Ltd., Toc- sin Analyser, kindly made available by the Wessex Water Authority. Rivers Division. was used for total organic carbon determinations.

5. RESULTS

5.1 Excitation and emissions spectra

Figure 1 presents spectra typical of those obtained from the samples. The excitation spectra are for an emission wavelength of 450 nm and the emission spec- tra are for an excitation wavelength of 350 nm. For all samples the peak excitation was at 340-350nm with a width of 20--30nm. A subsidiary peak was often evident at about 260-270nm, forming a shoulder on the curve. The width and relative

E x ~ t a t i o n E m i s s i o r ,

.o <

r,- ! i l

200 300 400 500

Wavelength (rim } Fig. 1. Typical excitation and emission spectra. Excitation spectra are for an emission wavelength of 450nm and emission spectra for an excitation wavelength of 350 nm.

Table 1. Percentage change in fluorescence after storage

Storage period Fluorescence relative to Day 1" t dayst ('I o)

3 +1.1 +5.3

9 -6.4

* Average for 24 samples,

strength of these peaks varied between samples, but both were usually present,

The emission spectra showed broad peaks in the 400 to 460 nm region. There was some evidence that peaks of a similar order of magnitude at 410-420 nm and 440-450 nm caused this form. These figures agree closely with the emission maxima found for a wide range of coloured waters in the United States by Ghassemi and Christman (1968) and with those reported by L'evesque [1972) for humic substances extracted from soils. The maxima at 500-540nm reported by Seal et al. (1964) were not observed. This may reflect the complex extraction and purification procedure used by these authors. The only samples which gave curves different to the spectral character- istics described above were a sample of rainwater, which showed a minor emission peak at 410 nm, and a sample of road runoff (B in Fig. 1) which showed peaks at 310 and 340rim in the excitation spectrum.

The Corning Cs 7-83 filter was selected to use the 365 nm line in the low pressure mercury lamp. This line is at a wavelength very near the peak excitation of the samples examined. A broader range of exci- tation wavelengths could be obtained using the Turner 110-850 general purpose lamp. which has a continuous emission in this region. Suitable filters would then be the Corning Cs 7-60 or Kodak 18A. Broad excitation and emission wavebands are desir- able because they reduce the selectivity of the analy- sis, The Kodak 64 secondary filter employed was chosen because of its broad transmission between 400 and 600rim. A Kodak 2A would be a satisfactory alternative, but use of a neutral density filter would probably be required. The filter combination used was found to be satisfactory for the range of samples examined, but further examination of filters and light sources would be worthwhile.

5.2 Sample stability

In order to assess sample stability with time, the fluorescence of the samples was taken after periods of 1, 3, 8 and 9 days. Although there was some vari- ation in the readings, this was not in a constant direc- tion (Table 1). A one-way analysis of variance between the different days readings proved that the differences were not significant at the 95°/; confidence limit. Repeat analyses of TOC were not made.

Fluoresence was also noted before and after filt- ration through the 0.45nm membrane filter. There was no significant decrease in fluorescence at the 95% confidence limits when the before and after filtration

808 P L. SMART, B. L. FINLAYsoN. W. D. RYLANDS and C, M. BALL

Table 2. Fluorescence and TOC after storage in dark cold and in light warm conditions

Stored in dark Stored in light Sample fluorescence TOC fluorescence TOC

M 82 6.3 80 " ~ L 216 13.4 252 16.2

data was compared using a one-way analysis of vari- ance.

Two samples were split after filtration and one sub- sample was stored under normal conditions, while the other was left in a sunny window at room tempera- ture. The fluoresence of sample L increased whilst that of sample M decreased (Table 2). A parallel in- crease in TOC was observed for sample L. but the TOC of sample M "also increased. In the case of sample L the parallel increase in fluorescence and TOC caused both subsamples to fall on the TOC/ fluoresence regression. For sample M the point moved away from the line. It is possible that this was because sample M was of road runoff.

These limited results suggest that sample instability will not cause significant deviations from any TOC/ fluorescence relationship. However, if the "'true" TOC of a water sample is required it would be more satis- factory to analyse it without a lengthy period of stor- age.

5.3 Temperature

Because fluorescence is temperature dependent it was necessary to analyse all samples at a standard temperature, 22°C in this case. However. it is some- times advantageous to analyse samples in the field. in which case a temperature correction will be required. Six samples were examined at temperatures from 13 to 35°C and the resulting curves fitted using an exponential regression of the form:

F = Foe" ' (1)

where F is the fluorescence at temperature t. F0 is fluorescence at 0~C and n is a constant temperature coefficient.

1.3

~'~C~ 1.2

¢.~ 0.9

0.8

• S a m p l e n("C - : ) r~ Fs=Fe m t s - t l D -o .o l8

J -o .m2

29 :o° 2,'o

124 d

I

u,. 0 7 i i i i | i B i J i i i i i -16 -14 -12 -~0 - 8 - 6 - 4 - 2 u 2 4 6 8 I0 12 14 16

Sample T e m p e r a t u r e D i f f e r e n c e F rom S t a n d a r d ( t s - t )

Fig. 2, Temperature correction curves.

Figure 2 presents correction curves derived from the temperature exponents and presented as:

F s = Fe'"*-" (2)

where F, is fluorescence at the standard temperature r,. It is apparent that the fluorescence is strongly dependent on temperature, and that the degree of this dependence varies from sample to sample. An average figure of -0.011':C - t was obtained, which is similar to the figure given by Christman and Minear (19671 for lignin sulphonates. It may be necessary to prepare specific curves if analysis at a standard temperature is not possible.

5.4 pH

The majority of natural waters have pH values between 6.5 and 8.5. Many samples, however, are likely to be more acid. for instance the highly col- oured waters draining peaty upland areas, pH/fluores- cence curves were therefore prepared for several samples using buffer solutions. Two types of curve were obtained (Fig. 3). one showed a fluorescence maximum at pH 5--6, but the other was sigmoid with little change in fluorescence between pH 6 and 8. These two types of response may also be identified in samples which were buffered to pH 4.0. 7.0 and 9.0 (Table 3). Christman and Arnquist (1969) and Christman and Minear (1967) both give pH;fluores- cence curves with maxima, while Black and Christ- man (1963) present data which probably represent the sigmoid response. Thus it is clear that pH has an effect on fluorescence, but between pH 5.0 and 8.0 these variations are relatively limited.

5.5 Other interference

Ghassemi and Christman (1968) and Christman and Arnquist (1969) have shown that there are com- plex interrelationships between organic fluorescence. pH and the concentration of iron Ill and aluminium

150

o r~

Io0

t;

,'7

"6 g al

~- so

/ /

/

2 3 ,4 5 6 7 8 9 10 ~2 13

B u f f e r e 0 Sample pH

Fig. 3. Relationships between pH and fluorescence.

Dissolved organic carbon in surface waters 809

Table 3. Fluorescence at pH 4.0 and 9.0 expressed as a percentage of fluorescence at pH 7.0

pH 4.0 pH 9.0 Probable type of curve

96 117 91 109

100 108 86 110 Sigmoid 96 108 90 109

83 94 102 88 92 93

104 94 Single maxima 97 89 98 89

105 102 Other

22

20

18

16

14

12

10

8

6 -

E 2 -

0 ~ 0

/ 1 : / "

! . / / . / / /

/ * / . / ' / / "

. / ~ . ~" / ,~l . / / / / / ' .~," IV

. ; - , : , - . / s ' ~, /

/ s . / l • f . i p ,,'

i IV' . , .-. , . ' /~,,.-

i " ~'11 .1 -

I I I I I 50 100

F l u o r e s c e n t I n t e n s i t y

2SE

I$E

-1SE

• 2SE

150 200 250 300

III. It was found that fluorescence was a good indica- tion of the optimum conditions for precipitation of organics from water, It has not been possible to measure the iron III or aluminium III concentrations of the waters sampled during the work reported here, thus the extent of any interference from this source is unknown.

No interference will be obtained from chlorophyll or any other common natural pigment. Chlorophyll fluoresces strongly in the far red part of the spectrum with a maxima at 665 nm (Yentsch and Menzel, 1963). Similarly fluorescent algae are very rare in temperate waters (Round, 1965). However, the presence of opti- cal brighteners could present problems. These highly fluorescent blue dyes are widely used in washing powders to "whiten whites". They have an emission maxima at 460nm which is very close to that reported for the waters used in this study. Fortunately these compounds bind very strongly to cellulose, and consequently will not persist in significant concen- trations in sewage or natural waters.

5.6 T OC/fluorescence relationships

Simple linear regressions were calculated for the three groups of samples described in Section 3. A regression was also made between absorbance at 254 nm and TOC for samples in Group 1. Table 4 presents the results of the regression analyses, Group 3 samples have been corrected for the dilution caused by the addition of buffer solution. For each regression the probability level of the F statistic is given.

Fig. 4. Regression of TOC on fluorescent intensity for samples in Group 1.

Figure 4 shows the relationship between TOC and fluorescence for the 24 samples of Group 1. Despite the variety of sample sources the statistical relation- ship is particularly good. There is a small positive intercept on the TOC axis. The correlation between TOC and absorbance is correspondingly good for these samples Gable 4).

The regression for Group 2 samples (Fig. 5) is not significant at the 95°0 level (0.10 > P > 0.05), prob- ably because of the heterogeneous nature of organic pollutants in motorway runoff. Furthermore there is a significant positive intercept on the TOC axis which indicates that some non-fluorescent carbon atoms are present. The presence of long-non-fluorescent carbon chains derived from oil deposited on the road surface (Anon, 1974) would account for both the poor rela- tionship obtained and the positive TOC intercept.

Group 3 samples also showed a highly significant correlation between fluorescence and TOC, though the correlation coefficient is lower than for Group 1. (Fig. 6). It is clear that there is considerable scatter about the line, especially for high TOC values. These high TOC samples are almost exclusively soil throughflow waters which exhibit strong yellow/ brown colouration. Such waters are commonly as- sociated with the presence of iron (Lamar, 1968) which has been shown by Ghassemi and Christman

Table 4. TOC/fluorescenee and TOC/absorbance regression data

Sample Intercept Correlation No. of Probability group TOC/fluorescence regressions (TOC) Gradient coefficient samples level of F

1. Various sources 0.815 0.065 0.98 24 0.001 2. Motorway runoff 0.574 0.049 0.46 17 0.10 3. East Twin Brook 1.549 0.085 0.87 34 0.001

TOC/absorbance regessions 1. Various sources 2.226 24.114 0.98 24 0.001

810 P.L. SMART. B. L. FINLAYSON. W. D. RYLANDS and C. M. BALL

12

H

10

9

8

7

6

5

a

3

2o

,,9"" 2SE

1 /

. / ,s" " " . 1SE

. / " t ' . I f " •

. t . / ~ "

" 1 / / " ~ ' / " _ 1 . t . / . t ' ~ ' ] -2SE

i ~ I t t 1 1 I

I0 20 30 40 50 60 70 80 90 100 Fluorescent In tens i ty

Fig. 5. Regression of TOC on fluorescent intensity for samples of Group "

(1968) to affect fluorescence in a complex interrela- tionship with pH. Certainly iron is mobile in the peaty pods•Is of the upper part of East Twin basin. It is possible that the buffering of these water samples affected the fluorescence by altering the iron/organic interaction. Further work is necessary to evaluate this problem. It was found that the scatter was not greatly reduced by separating peaty waters from those de- rived from brown earths. Unfortunately not enough samples were available from one site to test whether a site specific calibration would have shown reduced scatter. The intercept on the TOC axis in this regres- sion is not significantly different to zero at the 95°0 level.

32 -

3 0 -

28

26

24

2 2

2 0

t8

Z6

la

12

10

8

r- 4

2SE 1SE

/ - O ~ " dO ,," / ' 0 / ' / "

/ " . /

/ " / / , / , , . , / ' / 'o . / / / . / • / / / " • / . /

/ ' . / • , / ' / / . / / /

/ • / / .

/ o • 0 / / "* / ' /

/q~./ o . / . ,e' I ,~ I I I 1

50 100 150 200 250 300

Fluorescent Intensi ty

-1SE

-2SE

Fig. 6. Regression of TOC on fluorescent intensit3 for samples of Group 3.

6. CONCLUSION

The preliminary results presented here clearly show that fluorescence can be used as a predictor of TOC over a wide range of concentrations. That this statisti- cal relationship may be generally applicable is indi- cated by the strength of the statistical relationships obtained despite the very varied water samples uti- lized. It is possible that even better relationships may be obtained by the use of site specific curves particu- larly for locations where mixing of different waters has not occurred, e.g. soil throughflow sites. Further work is required to assess the variability of results for one site through time, and to thoroughly investi- gate the influence of metals and pH on organic fluor- escence.

Although it is unlikely that fluorescence measured over an appreciable band width will prove a signifi- cantly better predictor of TOC than absorption at a single wavelength, the fluorescence method may possess some advantages. Fluorescence is consider- ably less affected by sample turbidity than absorbance which would facilitate continuous monitoring. Fur- thermore, robust filter fluorometers are available which may be operated under field conditions in both individual sample and continuous flow modes. This may be a considerable advantage in situations where immediate analyses are required, for instance when identifying sources of agricultural pollutants.

Acknowledqements--The authors gratefully acknowledge the assistance of Mr. David Holmes. of the Department of Geography. University of Bristol, in the collection of samples. Dr. P. Hancock. Mr F. Sweeting and Mr. R. New- ton of the Wessex Water Authority. Rivers Division. gener- ously allowed us the use of their TOC Analyser. Dr. Julia James. Department of Chemistry. University of Sydney. Australia. and Dr. T. C. Atkinson, School of Environmen- tal Sciences, University of East Anglia, provided valuable criticism during the preparation of the manuscript.

REFERENCES

Abrams I. M. (19691 Removal of organics from water by synthetic resinous adsorbents. Chem. Engn# Proq. Syrup. Set. 65 (97t, 106--112.

American Public Health Association 11965) Standard Methods.lor the Examination of H'ater and I4"astewarer. APH.A. . New York. 769 pp.

Anon 11974) The slippery problem of motorway muck. New Scient. 64 1918), 106.

Baker C. D., Bartlett P. D.. Farr I. S. & Williams G. I. (19741 Improved methods for the measurement of dis- solved and particulate carbon in fresh water and their application to chalk streams. Freshwat. Biol. 4 (51, 467-481.

Baumgartner D. J.. Feldman M. H. & Gibbons C. U (1971) A procedure for tracing kraft mill effluent from an ocean outfall by constituent fluorescence. Watcr Res. 5. 533 544.

Black A. P. & Christman R. F. I1963) Characteristics of coloured surface waters. J. Am. liar. IVks Ass. 55 (61. 753-77Q.

Dissolved organic carbon in surface waters 811

Christman R. F. (1970) Chemical structures of colour pro- ducing organic substances in water, pp, 181 198 Sym- posium on Or qanic Matter in Natural 14?tters. Edited b 3 Hood D. W. University of Alaska. Institute of Marine Science. Occasional Publication No. 1.

Christman R. F. & Arnquist J. L. (1969) Fluroescence tech- niques in detection of organics in water. A.S.T.M. Special Technical Publication 448, 96-115.

Christman R. F. & Minear R. A. (1967) Fluorometric determination of lignin sulphonates. Trend En~plg (;nit'. 1/~'ash. 19 ( 1 ). 3-7.

Dienert F. (1910~ De la recherche des substances fluores- centes dans le controle de la sterilisation des eaux. C. r. hehd. Seanc. Acad. Sci.. Paris 150 (8). 487-488.

Dobbs R. A.. Wise R. H. & Dean R B. (1972) The use of ultraviolet absorbance for monitoring the total organic carbon content of water and wastewater. Water Res. 6. t173-1180.

Fischer S. G. & Likens G. E. (1972~ Stream ecosystem: organic energy budget. Biol. Sci. 22. 33-35.

Ghassemi M. & Christman R. F. (1968) Properties of the yellow organic acids of natural waters. Limnol. Oceanogr. 13. 583-597.

Goerlitz D. F. & Brown E. (1972) Methods for analysis of organic substances in water. Techniques ~l VCater Resources Investigations ()f the U,ited States Geological Survey, Chapter A3. Book 5. U.S. Govt. Printing Office, Washington.

Hall K. J. & Lee G. F. (1974) Molecular size and spectral characterization of organic matter in a meromictic lake. Water Res. 8. 239-251.

II'in N. P. & Orlov D. S. (1973) Photochemical destruction of humic acids. Soviet Soil Sci. 5 (11. 75 83.

Jones R. H. (1969) TOC analysis and its relationship to BOD and COD. Ent'ironmental Pollution Instrumen- tation. Edited b 3 R. L. Chapman. pp. 116-125. Society of America. New York.

Lamar W. L. (1968) Evaluation of organic colour and iron in natural surface waters. U.S.G.S. Protl Pap. 600-D. D24-29.

Lamar W. L. & Goerlitz D. F. (1966) Organic acids in naturally coloured surface waters. U.S.G.S. Wat, Supp. Pap. 1817-A. 17pp.

L'evesque M. (1972) Fluorescence and gel filtration of humic compounds. Soil Sci. 113. 346-353,

Ling Ong H.. Swanson E. V. & Bisque R. E. (1970) Natural

organic acids as agents of chemical weathering. U.S.G.S. Prq/i Pap. 70(1-C. C 130-C 137.

Manny B. A. & Wetzel R. G, (19731 Diurnal changes in dissolved organic and inorganic carbon and nitrogen in a hardwater stream. Freshwat. Biol. 3. 31-43.

Mattson J. S.. Smith C. A., Jones T. T.. Gerchakov S. M. & Epstein B. D. (19741 Continuous monitoring of dis- solved organic matter bx ultraviolet-visible photometry. Limnol. Oceanogr. 19 t3). 530- 535.

Mrkva M. (1975) Automatic u.v.--control system for evaluation of organic water pollution. Hater Res. 9, 587-58~.

Packham R. F. (19641 Studies of organic colour in natural waters. Proc. Soc. ~lat. Treat. Exam, 13, 316-334.

Round R. E. (1965) The Biolo,o,y o! the Alqae. 269 pp. Edward Arnold, London.

Seal B. K.. Ro3 K B. & Mukheriee S. K. (1964) Fluores- cence emission spectra and structure of humic and fulvic acids. J. Indian chem. Sac. 41 (3), 212-214.

Skopintsev B. A. & Krylova L. P. (1955) Optical properties of the aqueous humus present in the upper layers of inland waters. Gidrokhtm. Mater. 23 (31), 38-45.

Societ3 for Analytical Chemistry (1958) Recommended Methods.['or the Analysis t~f Trade Effluents. Society for Analytical Chemistry. London.

Sowden F. S. & Deuel H. (19611 Fractionation of fulvic acids from the B horizon of a podsol. Soil Sci. 91, 44--47.

U.S. Public Health Service (1961) Drinking water stan- dards. J. Am. War. Wks Ass. 53, 935.

Van Hall C. E., Safranko J. & Stanger V. A. (19631 Rapid combustion method for the determination of organic substances in aqueous solutions. Anal3t. Chem. 35, 315-319.

Viro P. J. (1953) Loss of nutrients and the natural nutrient balance of the soil in Finland. Communicationes lnstituti Forestalls FemTiae 42 I l). 50.

Walling D. E. (19751 Solute variation in small catchment streams: some comments. Trans. Inst. Br. Geoqr. 64. 141-147.

Wilander A.. Kvarnas H. & Lindell T. (1974) A modified t]uorometric method for measurement of lignin sul- fonates and its in situ application in natural waters, Water Res. 8, 1037-1045.

Yentsch C. S. & Menzel D, W. (19631 A method for the determination of phytoplankton chlorophyll and phaeo- phytin by fluorescence. Deep Sea Res. 10. 221-231.