Detection of Escherichia coli and Enterobacter aerogenes in water samples collected from two sites, one site near South Shore Water Treatment Facility, the other farther from the Facility, on Lake Michigan, in Milwaukee, Wisconsin

 

 

 

Stephani Duelge

Melissa Unruh

BI 341, Fall 2002

Research Report

November 2, 2002

 

Abstract

Water was collected from two sites in Milwaukee County on the shoreline of Lake Michigan to determine if there is a difference in the number fecal coliforms in the water.  The locations selected were located 3.24 km and 13.35 km upstream from the South Shore Water Treatment Facility.  The water was collected from the two sites on four different days.  The water was inoculated into lactose broth and observed for gas production and the amount of gas production was recorded.  The lactose broth was used to inoculate plates that are selective for fecal coliforms.  Inoculating lactose broth and nutrient agar slant with the plates containing the selective media was the way we performed a confirmatory test.  The broth was measured again for gas production and the nutrient agar slant was used to make gram stain slides to confirm the presence of gram-negative rods that also ferment lactose, or coliforms.  We found no significant difference between levels of coliforms between the two sites (p = 0.421). 

Keywords:  fecal coliforms, Lake Michigan, South Shore Water Treatment Facility, Escherichia coli, Enterobacter aerogenes

Introduction

            In Milwaukee, Wisconsin and the surrounding areas, much of the fresh water used in homes and industry originates from Lake Michigan.  However, Lake Michigan is also the dumping point for much of the wastewater.  In order to ensure that water is safe for human consumption, water is treated locally at the Jones Island and South Shore Water Treatment Facilities.  We hypothesized that water samples near to, and far from, the South Shore Water Treatment Facility will show different levels of fecal coliforms bacteria.

            Ideally, direct detection of pathogens gives the best measure of the prevalence of pathogens in water, however applications of these methods are limited by time, expense, and difficulty adapting microbiological techniques to field applications (Islam, et al, 2000).  Instead, a common practice, and the one used by the Milwaukee Metropolitan Sewerage District (MMSD), is to test for the presence of fecal coliform bacteria.  Fecal coliform bacteria are defined as facultative anaerobes that ferment lactose to form gas, and are Gram-negative, non-spore forming rods (Brion, Mao, 2000).  The two types of fecal coliform bacteria are Escherichia coli and Enterobacter aerogenes.  E. coli is considered a good sewage indicator because it is not normally encountered in water or soil, is easily identified, and does not form endospores, so it does not survive for extended periods of time in water (Islam, et al, 2000).  When it is found, it is safe to assume that water is unfit for consumption due to fecal contamination.  Also, tests for the nonpathogenic E. coli can imply that other pathogenic intestinal bacteria are present, when those pathogens may be difficult to identify from a water sample (Islam, et al, 2000).  Use of fecal coliforms as indicators has reduced the incidence of waterborne disease since the onset of water treatment in the early twentieth century (Brion, Mao, 2002).

            The purpose for this experiment was to determine if levels of fecal coliform bacteria were higher nearer to the South Shore Water Treatment Facility than levels away from the facility.  The Department of Natural Resources has set safe limits of fecal coliform and total coliform bacteria levels of 400 individuals per 100 mL of water (Shafer, 2001).

Methods

For this experiment, it was decided to test the numbers of fecal coliform bacteria in Lake Michigan on two sites.  The first site was near the South Shore Water Treatment Facility, in Oak Creek, Wisconsin.  The outlet pipe for the facility was 3.218 km from the actual plant.  The site chosen for sampling was 0.4023 km from the actual plant, due to restricted access to the water.  This made a straight-line distance from the outlet pipe of 3.243 km.  This was called the South Shore site in all future references.  The second site was 13.355 km from the actual facility, making the straight-line distance from the outlet 13.737 km.  This was called the Sheridan Park site in all future references.  Water samples were collected on three consecutive days, one from each site (see table 1).  A fourth sample was collected non-consecutively, due to sample contamination on Sunday, October 27, 2002.

Table 1: Locations, dates, and times of sample collection.

Site

Date Collected

Time Collected

South Shore

Thursday, October 24, 2002

12:15 p.m.

Sheridan Park

Thursday, October 24, 2002

12:42 p.m.

South Shore

Friday, October 25, 2002

11:44 am

Sheridan Park

Friday, October 25, 2002

12:22 p.m.

South Shore

Saturday, October 26, 2002

11:31 a.m.

Sheridan Park

Saturday, October 26, 2002

12:09 p.m.

South Shore

Monday, October 28, 2002

5:04 p.m.

Sheridan Park

Monday, October 28, 2002

5:42 p.m.

 

 

            For collection, 200 mL glass bottles with caps were sterilized with caps on by autoclaving at 121˚C for 15 minutes.  Sample bottles were fully submerged in the lake water before caps were removed.  Water was allowed to fill the bottle, expelling any air left in the bottle before caps were replaced.  Only after the cap was tightened was the bottle removed from the water.  By collecting samples aseptically, the chance of sample contamination from the non-water environment was decreased greatly.  Because the water samples collected from Lake Michigan were non-potable, samples needed to be tested immediately after collection.  When this was not possible, samples could be stored for 8 hours at 8˚C.  However, in this procedure, all samples were cultured within a 2-hour time period.

            In preparation for the cultures the media was prepared as follows:

Single Strength Lactose Broth (SSLB):

17 grams of dehydrated lactose broth media was dissolved in 500 mL of water.  From this, 100 culture tubes with inverted Durham tubes were filled with 5 mL of the broth.  All tubes were capped and autoclaved at 121˚ C for 15 minutes.  Tubes were refrigerated to prevent contamination.

 

 

Double Strength Lactose Broth (DSLB):

8.5 grams of dehydrated lactose broth media was dissolved in 125 mL of water.  From this, 25 culture tubes with inverted Durham tubes were filled with 5 mL of the broth.  All tubes were capped and autoclaved at 121˚ C for 15 minutes.  Tubes were refrigerated to prevent contamination.

Endo Agar Plates:

14 g of dehydrated Endo agar media was dissolved in 500 mL of water and brought to a rolling boil.  The media was then autoclaved at 121˚C for 15 minutes.  20 mL of the agar was poured into Petri plates that were sterilized under UV light and allowed to cool.  The Endo agar plates were capped and stored upside down in the refrigerator for future use.

Levine EMB Agar Plates:

15.5 g of dehydrated Levine EMB agar media was dissolved in 500 mL of water and brought to a rolling boil.  The media was then autoclaved at 121˚C for 15 minutes.  20 mL of the agar was poured into Petri plates that were sterilized under UV light and allowed to cool.  The Levine EMB agar plates were capped and stored upside down in the refrigerator for future use.

Nutrient Agar Slants:

14.6 g of dehydrated nutrient agar media was dissolved in 400 mL of water and brought to a rapid boil.  8 mL of the agar was poured into culture tubes, and the tubes were capped.  The media was then autoclaved at 121˚C for 15 minutes.  The tubes were promptly removed from the autoclave and tipped on the side to form a slanted surface for bacterial growth.  The slants were refrigerated for future use.

            Immediately after the collection of the samples were brought back to Alverno College for culturing.  From each sample, three sterile DSLB tubes were inoculated with 10.0 mL of the sample, three sterile SSLB tubes were inoculated with 1.0 mL of the sample, and three sterile SSLB tubes were inoculated with 0.1 mL of the sample.  All pipettes used in transfer were sterile, and all work was done in the sterile UV hood. 

All tubes were then incubated at 34˚C for 48 hours.  After 48 hours, all tubes were examined for gas production, a sign that lactose-fermenting bacteria were present.  The inverted tubes were 5 cm long, and to determine the percent gas produced, the gas bubble was measured, and then the measured length was divided by 5 cm and then multiplied by 100.  When 10% or more gas was present in two or more tubes, the water was presumed to contain levels of fecal coliforms unfit for human consumption.

            From this data, the number of bacteria present in each sample was determined.  The most probable number (MPN) of coliform bacteria present in 100 mL of water was decided using Appendix A, Table VI, “MPN Determination from Multiple Tube Test,” (Benson, 2002).  This table gives the 95% confidence that the MPN is as shown in table 10 in the results section.

            After determination of the MPN, the tube showing the highest percentage of growth was chosen for inoculation of Endo agar plates and Levine EMB agar plates.  Endo agar contains a fuschin-sulfite indicator.  When fecal coliforms grow on Endo agar plates, colonies appear red, while other bacterial type colonies remain colorless.  Levine EMB agar contains methylene blue, which inhibits the growth of gram-positive bacteria.  Coliform bacteria form nucleated colonies on the media, and E. coli colonies exhibit a greenish-metallic sheen. 


 


Figure 1:  Growth on a Levine EMB agar plate.  The shiny metallic green colonies show the presence of E. coli.

Figure 2:  Growth on an Endo agar plate.  The red colonies and the absence of red color of he agar show the presence of gram-negative fecal coliforms.

 

 


Plates were inoculated by streaking the plate in a quadrant method, to provide good isolation of colonies.  All plates were then incubated at 34˚C for 24 hours.  After 24 hours, the plates were checked for coliform growth. 

            After plates were examined for coliform growth, one sterile SSLB tube, and one sterile nutrient agar slant was inoculated from a coliform colony on each plate.  The SSLB Durham tubes and nutrient agar slants were incubated at 34˚C for 24 hours.

            After 24 hours, the SSLB Durham tubes were examined for gas production, as a double check for the presence of coliform bacteria.  However, the absence of gas does not rule out the presence of coliforms since the tubes must incubate for 48 hours before a negative result can be confirmed.       Also, after 24 hours, slides were prepared using the growth from the nutrient agar slants.  First, a drop of distilled water was placed on a clean slide, and one loopful of bacteria was placed in the water and allowed to air dry.  After drying, the smear was heat fixed with a Bunsen burner.  Because the confirmation of fecal coliforms requires the identification of gram-negative rods, smears were gram-stained and examined under a microscope

            This was the final confirmation stage for this experiment.  All tubes and plates were autoclaved to kill any remaining live bacteria and cleaned and disposed of properly.

Results

Tables 2-9 show the amount of gas formed by lactose fermenting bacteria in the Durham tubes.  In order for the tube to be positive for lactose fermenting bacteria, 10%, or more gas must have been produced within 48 hours from inoculation.  Because we were interested in comparing two sites, even the tubes that displayed less than 10% gas formation were considered in subsequent comparisons.


 

Day 1

 

Day 1

South Shore

 

Sheridan Park

Broth

Tube 1

Tube 2

Tube 3

 

Broth

Tube 1

Tube 2

Tube 3

DSLB 10mL

52%

56%

58%

 

DSLB 10mL

100%

80%

62%

SSLB 1.0mL

22%

14%

18%

 

SSLB 1.0mL

70%

30%

22%

SSLB 0.1mL

0%

0%

22%

 

SSLB 0.1mL

12%

24%

16%

Table 2

 

 

 

 

Table 3

 

 

 

 

Day 2

 

Day 2

South Shore

 

Sheridan Park

Broth

Tube 1

Tube 2

Tube 3

 

Broth

Tube 1

Tube 2

Tube 3

DSLB 10mL

28%

28%

32%

 

DSLB 10mL

40%

40%

32%

SSLB 1.0mL

8%

10%

36%

 

SSLB 1.0mL

12%

30%

32%

SSLB 0.1mL

2%

8%

56%

 

SSLB 0.1mL

38%

8%

6%

Table 4

 

 

 

 

Table 5

 

 

 

 

Day 3

 

Day 3

South Shore

 

Sheridan Park

Broth

Tube 1

Tube 2

Tube 3

 

Broth

Tube 1

Tube 2

Tube 3

DSLB 10mL

6%

34%

24%

 

DSLB 10mL

0%

16%

26%

SSLB 1.0mL

2%

0%

0%

 

SSLB 1.0mL

0%

0%

0%

SSLB 0.1mL

0%

0%

0%

 

SSLB 0.1mL

0%

0%

0%

Table 6

 

 

 

 

Table 7

 

 

 

 

Day 4

 

Day 4

South Shore

 

Sheridan Park

Broth

Tube 1

Tube 2

Tube 3

 

Broth

Tube 1

Tube 2

Tube 3

DSLB 10mL

52%

22%

28%

 

DSLB 10mL

20%

78%

50%

SSLB 1.0mL

12%

12%

12%

 

SSLB 1.0mL

16%

8%

26%

SSLB 0.1mL

100%

0%

2%

 

SSLB 0.1mL

10%

6%

8%

Table 8

 

 

 

 

Table 9

 

 

 

 

 

Table 10

 

The above table (10) shows the MPN, determined in the manner presented in the methods section of this report, is listed for each site, by the day the sample was collected.  Day 3 for both sites shows a significantly reduced number of coliforms present per 100 mL of water compared to the other three collection dates.

Figure 3: Bar graph showing the MPN at both sites on days 1-4.

            This graph (figure 3) shows the difference between the most probable numbers of coliforms present at both sites on all days samples were collected.  Note that the greatest difference was seen between days, not between sites.  The t-test results calculated to determine the significance of the MPN between both sites is 0.421.

Day 1

 

Day 1

Growth on Endo and Levine Agar Plates

 

Growth on Endo and Levine Agar Plates

South Shore

 

Sheridan Park

 

Endo Plate

Levine Plate

 

 

Endo Plate

Levine Plate

DSLB 10 mL

positive

positive

 

DSLB 10 mL

positive

positive

SSLB 1.0 mL

positive

negative

 

SSLB 1.0 mL

positive

positive

SSLB 0.1 mL

positive

positive

 

SSLB 0.1 mL

positive

positive

Table 11

 

 

 

Table 12

 

 

 

Day 2

 

Day 2

Growth on Endo and Levine Agar Plates

 

Growth on Endo and Levine Agar Plates

South Shore

 

Sheridan Park

 

Endo Plate

Levine Plate

 

 

Endo Plate

Levine Plate

DSLB 10 mL

positive

positive

 

DSLB 10 mL

positive

negative

SSLB 1.0 mL

positive

positive

 

SSLB 1.0 mL

positive

negative

SSLB 0.1 mL

positive

positive

 

SSLB 0.1 mL

positive

negative

Table 13

 

 

 

Table 14

 

 

 

Day 3

 

Day 3

Growth on Endo and Levine Agar Plates

 

Growth on Endo and Levine Agar Plates

South Shore

 

Sheridan Park

 

Endo Plate

Levine Plate

 

 

Endo Plate

Levine Plate

DSLB 10 mL

negative

positive

 

DSLB 10 mL

positive

positive

SSLB 1.0 mL

negative

positive

 

SSLB 1.0 mL

negative

positive

SSLB 0.1 mL

negative

negative

 

SSLB 0.1 mL

negative

negative

Table 15

 

 

 

Table 16

 

 

 

           

 

Day 4

 

Day 4

Growth on Endo and Levine Agar Plates

 

Growth on Endo and Levine Agar Plates

South Shore

 

Sheridan Park

 

Endo Plate

Levine Plate

 

 

Endo Plate

Levine Plate

DSLB 10 mL

positive

positive

 

DSLB 10 mL

positive

positive

SSLB 1.0 mL

positive

positive

 

SSLB 1.0 mL

positive

positive

SSLB 0.1 mL

positive

positive

 

SSLB 0.1 mL

positive

positive

Table 17

 

 

 

Table 18

 

 

 

In the above tables (11-18) the results of coliform growth are listed by collection date and by site.  If there was red growth on the Endo agar plates, it was recorded that the plate was positive for coliform presence.  If no red growth was seen, it was recorded that the plate was negative for coliform presence.  Note that on day three, there were five Endo agar plates that were negative for coliform presence.  Note on day 2, for the Sheridan Park site, that all Levine EMB agar plates were negative for coliform production.

Day 1

South Shore

Broth That Media Was Inoculated From

Endo Broth

Endo Slant

Levine Broth

Levine Slant

DSLB 10mL

0%

gram (-) cocci

24%

gram (-) cocci

SSLB 1.0mL

20%

gram (-) cocci and rods

22%

gram (-) cocci

SSLB 0.1mL

20%

gram (+) rods, gram(-) rods

32%

gram (+) rods and cocci

Table 19

 

            This table (19) shows that there were gram-negative rods present, which indicate the presence of coliforms.  When considered in conjunction with the positive results listed for the growth on the agar plates, this indicates that coliforms were present in the MPN determined for day 1 at the South Shore site of 460 bacteria per 100 mL of water. 

Day 2

South Shore

Broth That Media Was Inoculated From

Endo Broth

Endo Slant

Levine Broth

Levine Slant

DSLB 10mL

6%

Gram (+) cocci

20%

gram (-) cocci and gram (+) cocci

SSLB 1.0mL

26%

gram (-) rods

14%

gram (-) rods

SSLB 0.1mL

62%

Gram (+) rods

52%

gram (-) rods and cocci

Table 20

            This table (20) shows that there were gram-negative rods present, which indicate the presence of coliforms.  When considered in conjunction with the positive results listed for the growth on the agar plates, this indicates that coliforms were present in the MPN determined for day 2 at the South Shore site of greater than 1100 bacteria per 100 mL of water. 

Day 3

South Shore

Broth That Media Was Inoculated From

Endo Broth

Endo Slant

Levine Broth

Levine Slant

DSLB 10mL

6%

gram (+) cocci and rods

0%

gram (+) cocci

SSLB 1.0mL

0%

gram (+) cocci and rods

0%

gram (+) rods

SSLB 0.1mL

0%

Gram (+) rods

0%

gram (-) cocci

Table 21

                        This table (21) shows that there were no gram-negative rods present, which indicate the absence of coliforms.  When considered in conjunction with the negative results listed for the growth on the agar plates, this indicates that coliforms were not present in the MPN determined for day 3 at the South Shore site of 43 bacteria per 100 mL of water. 

Day 4

South Shore

Broth That Media Was Inoculated From

Endo Broth

Endo Slant

Levine Broth

Levine Slant

DSLB 10mL

22%

Gram (+) cocci

14%

gram (+) and gram (-) rods

SSLB 1.0mL

30%

Gram (-) rods

0%

gram (-) rods

SSLB 0.1mL

2%

Gram (-) rods

0%

gram (-) rods

Table 22

                        This table (22) shows that there were gram-negative rods present, which indicate the presence of coliforms.  When considered in conjunction with the positive results listed for the growth on the agar plates, this indicates that coliforms were present in the MPN determined for day 4 at the South Shore site of greater than 1100 bacteria per 100 mL of water. 

Day 1

Sheridan Park

Broth That Media Was Inoculated From

Endo Broth

Endo Slant

Levine Broth

Levine Slant

DSLB 10mL

16%

gram (-) rods and gram (+) rods

20%

gram (-) rods

SSLB 1.0mL

24%

gram (-) rods

36%

gram (-) rods

SSLB 0.1mL

16%

gram (-) cocci

0%

no growth

Table 23

                        This table (23) shows that there were gram-negative rods present, which indicate the presence of coliforms.  When considered in conjunction with the positive results listed for the growth on the agar plates, this indicates that coliforms were present in the MPN determined for day 1 at the Sheridan Park site of greater than 1100 bacteria per 100 mL of water. 


 

Day 2

Sheridan Park

Broth That Media Was Inoculated From

Endo Broth

Endo Slant

Levine Broth

Levine Slant

DSLB 10mL

12%

gram (-) cocci

20%

gram (+) rods

SSLB 1.0mL

52%

gram (+) rods

24%

gram (+) rods

SSLB 0.1mL

8%

gram (+) rods

0%

gram (+) rods

Table 24

                        This table (24) shows that there were gram-negative rods present, which indicate the presence of coliforms.  When considered in conjunction with the positive results listed for the growth on the agar plates, this indicates that coliforms were present in the MPN determined for day 2 at the Sheridan Park site of greater than 1100 bacteria per 100 mL of water. 

Day 3

Sheridan Park

Broth That Media Was Inoculated From

Endo Broth

Endo Slant

Levine Broth

Levine Slant

DSLB 10mL

32%

gram (+) cocci

52%

gram (+) cocci

SSLB 1.0mL

0%

gram (+) cocci

0%

gram (+) cocci

SSLB 0.1mL

0%

gram (+) cocci

0%

gram (+) cocci

Table 25

            This table (25) shows that there were no gram-negative rods present, which indicate the absence of coliforms.  When considered in conjunction with the negative results listed for the growth on the agar plates, this indicates that coliforms were not present in the MPN determined for day 3 at the Sheridan Park site of 9 bacteria per 100 mL of water. 

 

 

Day 4

Sheridan Park

Broth That Media Was Inoculated From

Endo Broth

Endo Slant

Levine Broth

Levine Slant

DSLB 10mL

18%

gram (-) rods

0%

gram (-) rods

SSLB 1.0mL

42%

gram (-) rods

20%

gram (-) rods

SSLB 0.1mL

2%

gram (-) rods

0%

gram (+) cocci and gram (-) rods

Table 26

            This table (26) shows that there were gram-negative rods present, which indicate the presence of coliforms.  When considered in conjunction with the positive results listed for the growth on the agar plates, this indicates that coliforms were present in the MPN determined for day 4 at the Sheridan Park site of greater than 1100 bacteria per 100 mL of water. 


Discussion

            This procedure was labor intensive and could have been performed in a time and cost efficient manner to determine the numbers of coliforms present per 100 mL of water.  Due to lack of materials and time constraints, the present method was used.

However, there were drawbacks.  There was the fact that when selecting media to inoculate from, there was room for human error.  We were consistent in choosing the broth or plate that showed the most growth in a 48-hour period, however, in the case of the  lactose broth tubes, there was a chance that the lactose fermenting bacteria present was not a coliform.  For instance, Clostridium perfringens is a gram-positive lactose fermenter (Brion, Mao, 2002).  The tube with the largest percentage of gas growth was not necessarily the tube with the largest composition of coliforms.

Still, the tube with the most gas produced was chosen to inoculate the Endo and Levine EMB agar plates.  Levine EMB agar inhibits the growth of gram-positive bacteria.  This indicates that the only bacteria that can grow on the media are gram-negative, and because they were streaked using the tube with the most gas present, the bacteria that grow will also be lactose fermenters (Islam, et al, 2000).  Proteus vulgaris, or another gram-negative lactose fermenter would grow on the media, even though it is not a coliform (Brion, Mao, 2002).  This might account for the positive result on Levine EMB agar on day three, even though no gram-negative rods were observed under microscopic examination.

The Endo agar contains a fuschin-sulfite indicator, which causes coliform colonies to appear red, due to their ability to utilize the carbon from the compound.  On day three, no coliform colonies were observed under microscopic examination, however one Endo agar plate showed positive for growth of coliforms.  This occurrence could have been the result of a septic contamination from the inoculating loop, or the presence of a coliform.  Because none of the other tests provide evidence that coliforms were present in the sample, we determined that because it was the first plate streaked from the grouping of plates, the loop had a chance of being contaminated.

On the other hand, there were incidences where gram-negative rods were seen under microscopic examination, yet plates showed no coliform growth.  This was again, a problem of human judgment.  Some of the plates were so overgrown with bacteria, that it was hard to differentiate between colonies, or to determine if colonies of bacteria had grown over the coliform colonies.   

On the third day, water samples were collected, and while the temperature of the lake water was a constant 8.5˚C on every day at each collection site, there was a noticeable decrease in the water movement.  On days 1, 2, and 4, wave height was high enough to splash  up to the thighs of the person collecting the sample.  However, on day 3, the waves were small enough that the person collecting the sample was able to stand in the water and walk out dry.  This might account for the fact that so few bacteria of any kind were present in the samples from day 3.

Comparatively, there was not a significant difference between coliform levels at the Sheridan Park site and the South Shore site (p = 0.421).  However, there was a notable difference between our results and the average results of 31 coliform per 100 mL of water, collected by MMSD during the year 2001.  As we have shown, there are differences in levels from day to day.  Also consider that MMSD samples a much larger region containing many more sites.  In order to sample such a large lake, the methods involved need to be much more efficient than the multiple tube method used here.  A faster method, performed on many more samples, would yield more accurate and precise results.

            If we were to repeat this experiment we would procure m-ColiBlue24 Broth to perform the experiment.  The m-ColiBlue24 method is less labor intensive so more sites could be tested since less time in the lab testing samples would be required.  This method, since it is simpler, would be more accurate than the method we used because there is less room for error.

 

Literature Cited

 

Benson, Harold.  2002.  Microbiological Applications:  Laboratory Manual in General Microbiology.  New York; McGraw Hill.  338 pp.

 

Brion Gail, Mao, Huazhong.  2002.  Use of Total Coliform Test for Watershed Monitoring with Respect to Atypicals.  Journal of Environmental Engineering, 126:  175-181.

 

Islam, M., Begum, A., Khan, S., Sadique, M., Khan, M., Albert, M., Yunus, M., Huq, A., and Colwell, R.  2001.  Microbiology of Pond Ecosystems in Rural Bangladesh:  Its Public Health Implications.  International Journal of Environmental Studies, 58:  33-46.