Three acidic forests

The title looks simple, but at the same time it’s a difficult one. Of course I haven’t been looking at the trees and other plants, because it’s all about the soil. Then I wondered whether I should talk about woods or forests. To me a limited area with trees won’t be qualified as a forest, but the FAO seems to be quite tolerant.

Some time ago I did several experiments with soil from the Spanderswoud (near Bussum), which turned out to be very acidic (pH 4 or less). This time I visited the forest to the North of Austerlitz in the Netherlands, a modest wood near Winterswijk, close to the river Slinge and a spot at the Holterberg. To get an impression about the soil I collected samples from an area of about ten metres along a path (actually two different ones for Austerlitz) and combined them in a bottle. Again the samples were superficial – between 5 and 10 cm deep. Later on we’ll probably look at deeper soil.

Photos are only meant to give an impression – not taken at the sampling locations
(left one manipulated, because people could be recognised).

Although the woods in Almere (formally also forests) were rather basic, this was probably because the soil is heavy clay. The three locations investigated this time, all had sandy soil mixed with humus. The top layer of dead organic material was scraped of before sampling, but still a lot of organic material was in. As learnt from previous experiments the soil was air-dried first and investigated without sieving (except for removal of stones and large pieces of wood).

Spoiler alert: all pH values were below 4!

For each sample-set (remember that between five and ten samples over a distance of ten metres were combined in a single bottle) two slurries were made out of 30 grams of soil with 30 grams of demineralised water.

Although sometimes the slurry was really compact, it was possible to measure both pH and conductivity. Each slurry was measure three times. The first round was for the pH, then one for conductivity and so on. Taking two slurries is a necessary approach, because not all the samples were very homogeneous. It’s not easy to have the same balanced combination the sandy and organic fraction every time. For the pH this is not very important, it seems, but the conductivity is very sensitive to different compositions. It was really interesting to see that a rather solid slurry could have a higher conductivity than a more fluid one.

This time I will present the actual values measured. The pH meter was checked with the buffers before and after. If the deviation became too large (e.g. over 0.05), a calibration was performed (beforehand of course, not afterwards). Usually some drift is observed – also depending on the temperature – but the values stay close to the real values, so the measurements are reliable.

Overview of all single measurements and some derived values

The Spanderswoud is left out, because there we used a 1:2 mixture of soil and water and the measurements were only done twice. The latter is not an issue, but different dilutions will lead to different values for both pH and conductivity.

Doing statistics, there is a complication. The three measurements for the same slurry will be correlated closely, but the two slurries derived from a single sample-set are correlated as well. We can’t simply assume that we are dealing with six different observations of equal weight. Yet the pH values were very close together – even so close that we present an average, minimum and maximum for the total values of all the locations. The individual values are also shown in a bar-graph.

All single results for pH combined in a graph

For the conductivity the pattern is less consistent. When a sample-set has both sandy soil and organic matter it is hard to take a representative sample from it. As a result the two slurries made out of the same sample-set got very different conductivities. Then different attempts to determine the conductivity in a single slurry can also differ a lot. This can be seen very clearly in the bar-graph.

All single results for conductivity (µS/cm) combined in a graph

Stepping back, the conclusion is that the three locations have a very low pH, meaning a lot of acid is present. This is something the Netherlands are struggling with indeed. The conductivity is less uniform, meaning the content of soluble salts is quite different for the locations. A lot of nettles and blackberries were present, especially in Winterswijk, near the Slinge.


Non-agricultural soil as a reference

Previously a series of samples were taken from a field near the “van Wagtendonkpad” in Almere. It is a nice circumstance that the field lays in between two recreational woods. Although some forestry will be done, those areas are not intensively managed, so we have nice references!

We already know that the field was rather homogeneous (apart from one strip with elevated conductivity). Therefore sampling only at the start and the end of the field would do. In the end four strips were sampled. The strips were about 200 metres long and every 20-50 metres superficial samples (5-10 cm, because it was all heavy clay) were taken and combined in a single bottle per strip. Thus four bottles with mixed samples were obtained. For locations 3 and 4 the sampling was done both to left and right of the path (alternating, at least 1 metre from the path).

The way of working for the analysis was rather similar to the approach for the agricultural field. First the samples were spread out on cardboard and left to dry. Larger lumps were crushed. After 2 hours, when most of the water was gone, the first slurries were prepared by taking 20 gram of soil and mixing it with 20 gram of (demineralised) water.

Only the addition of CaCl2 was left out, because it only seems to lower the pH and for conductivity measurements it isn’t useful at all.

After additional drying, 24 hours after sampling, the remaining soil samples were crushed another time and sieved (again coarse net, openings about 2 mm). Then the procedure was repeated, mixing 20 grams of soil with an equal weight of water.

Because weighing was done at every step, the percentage of lost water could be determined after 2 and 24 hours. It varied for the locations, but in the first step the weight went down between 11% and 19% and after the second step the (calculated) total loss of weight was between 15% and 24%.

In the end the sieving removed about 1/3rd of the weight (lumps, shells, pieces of wood). To be sure the removed parts did not have a completely different composition, the discarded fractions were put together, mixed thoroughly and then a single sample of 30 grams was taken out. After mixing with an equal amount of water, both pH and conductivity were measured several times.

Now it’s time for the results and some conclusions. This time, because we are looking at the non-agricultural areas, the reference is the agricultural field (next to the van Wagtendonkpad – the Lawsonpad field is ignored).

Let’s have a look at the pH first.

the order of the locations was changed, because 1 and 4 are close to each other, as are 2 and 3

Of course we can’t tell the significance from the graph, but the Standard Error of the Mean for the pH of the agricultural samples is very low, looking at the measurements made after 2 hours of drying and without sieving. Therefore the difference with even the closest value from the non-agricultural sample-set is still significant (99%).

For the additionally dried and sieved samples, the pH did not change a lot and the differences are still significant – even for the closest values. It’s safe to say that the agricultural soil is slightly more basic, but despite the significance it doesn’t seem too relevant. It can’t be the shells, because those were present in all the soil everywhere. By the way: the order of the locations was changed, because 1 and 4 are close to each other, as are 2 and 3 (see map).

Then it’s the conductivity’s turn.

We can clearly see that the conductivity is not the same for the four locations. It’s interesting to see that the higher value (location 2), is close to the place where the samples in the field also showed a higher conductivity (next to Homeruspad). Perhaps there is a strip of land with different properties running through both the agricultural and non-agricultural area?

Mainly because of the outliers, the differences are not really significant, although in general the conductivity is higher for the non-agricultural soil. That’s not what we expected! Is the soil exhausted after the harvest and waiting for fertiliser?

Rather than doing all kinds of statistical tricks, it’s interesting to put all conductivity observations together and sort them (basically a  non-parametrical approach). The result is shown below.

It’s very remarkable that the order flips after 24 hours and the suspicion is that this could be caused by the sieving. For the non-agricultural samples the discarded fraction was kept and measured as well. The conductivity of this slurry is much lower, but (knowing that about 1/3rd was kept out) the weighted average of the sieved fractions and the fractions to be discarded is very close to the original overall average! In the end there is a risk that the conducting materials are kept, whilst discarding the fraction with a lower conductivity. The real risk is that the split won’t be done equally for all kinds of samples.

Drying is inevitable, because without drying it’s nearly impossible to get a homogeneous mix of the sub-samples (unless it is all turned into slurry). Yet sieving is a risk, especially when a relevant part of the total sample is being kept out. Sieving the slurry could be a good compromise.

For now the conclusion is that drying is OK, but sieving of dry material is not. The CaCl2 doesn’t seem to add value at all. Let’s see where the new approach will bring us!


An unplanned river.

In the past I used to travel to the Hague five days a week (for a previous assgnment), but since the lockdowns, working from home became the standard. Nowadays I travel to the Hague only once a week and I thought it would be nice to take advantage of it for water sampling. Looking around at the map I spotted two lakes, near Voorschoten: Vlietland and Starrevaart.

Entering the road next to the lakes in my navigation system, I expected to get close to the water, but unfortunately the road was cut in two disconnected parts and I ended up in the wrong part. There was no time for a walk and sample for half an hour at least, so I went back to my car, but then I realised that the (bicycle) bridge I crossed went over the river “the Vliet”.

It’s not a matter of principle that I would stick to lentic systems (standing water like lakes) and ignore the lotic ones (streams, like rivers). It was just that the lakes were easily spotted at the map whereas the Vliet was hardly visible and I really had to look for it. It’s a small river, but worth sampling and so I did. This time I took four samples around the bridge, kept them and did the measurements in the car again. In between the series a boat passed by, probably mixing the water a little bit.

Blue circles indicate the sampling locations

Usually I present graphs, but a table is shown first. Values for the samples are pretty close. The systematic change in conductivity is not exceptional. Perhaps a small particle was covering a part of an electrode, changing the values by 1%.

The averages per measuring round are to check whether a systematic error can be observed. Averages per sample (only two measurements, but used to check unexpected differences between the samples) are not very informative and left out.

The graph won’t be very exciting, so this is a good opportunity to compare a couple of locations and also see what the behaviour of the samples is over time (kept in the dark at – still rather high – room temperature).

Earlier, I noticed that the pH will be lower after a couple of days, with the conductivity being increased. Then, by accident, I forgot about some samples (Blocq van Cuffeler, IJmeer) only to rediscover them nearly two weeks later. For the Merwede and the Vliet I thought one week would be nice. The graphs presented below cannot be compared without some hesitation, because for some locations several samples were kept (Merwede, Vliet) and for others it was only the third one, but it’s still informative in a qualitative way.

As expected from earlier observations, the pH goes down and the conductivity increases after keeping the samples for a week.

The pH goes down but the conductivity is quite stable. It’s only five days

The forgotten one, hence a long period (13 days) and a moment in between. Again pH going down, conductivity getting higher.
pH going down, like for all the others, but conductivity decreasing as well.

In the end it is clear that the pH will get lower over time (under the conditions as stated). Other observations showed the same trend, but the ones above were combined with conductivity.

Initial pH values were between 8 and 8.5 and conductivity is often close to 1000 µS/cm, except for the Merwede. Probably this is because of the large volume of fresh water coming in from its source, containing less soluble salts.


Dried sieved and combined soil samples – two days later

Two days later, crushing the remaining lumps again after an additional drying period, I sieved the dry soil and combined the samples of the “van Wagtendonk” field, because of the similarity of the values (the five samples themselves were already a combination of ten sub-samples each, taken along the sampling strip).

From the combination, 20 gram of soil was remixed with an equal amount of (demineralised) water and measured again. The same was done for the “Lawson” potato-field.

As usual duplicate measurements were taken, but because of the huge differences between the first and second values for conductivity (and then the change being even in different directions for the two fields), six measurements were taken this time – all about five minutes apart. This could reveal a time-related shift, if present. For completeness the pH values were determined four additional times as well, although no surprises were expected there.

The measurements were alternated for the two fields and the instruments were rinsed and dried between every measurement. In between the samples were stirred. As always the pH-meter was checked with the buffers, to be sure the values were not more off than a couple of hundreds (if so, calibration would be performed).

The measurements 3, 4 5 and 6 for conductivity mainly confirmed the second one, making the first one an outlier. The values for the “Lawson” field were actually close to the ones observed in the previous session. The “van Wagtendonk” field however, showed much higher conductivity values!

The pH was not too different from the range observed earlier (just a little bit lower), but the conductivity had changed a lot for the “van Wagtendonk” field. The “Lawson” potato-field was more or less unchanged. Of course the additional drying may have increased the concentration of salts present, but this can’t explain the 60% increase, because the humidity of the soil was rather low in the first place (as will be quantified in another post).

Several explanations are possible. It is unlikely that the fraction kept out during sieving contained less salt, but it is possible that particles present in there (and present in the slurry created with the un-sieved samples lowered the conductivity). It’s also possible that the finer particles will release the salt easier than the larger lumps. Of course one of the original samples had a high value for conductivity, but even this value was about 20% lower than the combined values.

In the end the real question is whether the rather high conductivity (even the values for the first round before sieving were ten times higher than the values of the previous samples, taken from a forest) is specific for the agricultural soil. So what we should do is take samples from the neighbouring wood and see if there is any difference.


Agricultural soil (at the bottom of a former sea)

Before presenting additional water samples, I really wanted to present another set of soil samples. The very acidic soil (pH as low as 3.5) in the wood (Spanderswoud, near Hilversum) was a surprise and I wanted to check the agricultural soil in our “new land”. New land basically means that it’s the bottom of a former sea. When the “polder” was created by pumping the water out of the area, the salt had already been washed out, because the “Afsluitdijk” was put in place first, changing the former sea into a freshwater lake. Therefore I didn’t expect a high conductivity as a result of the seawater once flowing here, although fertiliser might be influential.

In Almere there is a very elongated field next to the “van Wagtendonkpad”. The field is over 1.5 km in length, but rather narrow (about 25 metres wide). At one place it is suddenly broader, nearly 90 metres (I measured the sizes at Google maps to be sure).

Because of the very long stretch it seemed a nice field to investigate. Most likely it belongs to one farmer of at least it’s worked by the only one farmer. Previously onions were bred. I decided to sample at start, end and some spots in between, where paths were crossing. The land was freshly ploughed by the look of it and looked like heavy clay, with lumps of half a metre! The sample drill was of no use, because I could get it in only 10 cm and then the sample would not come out, so I started collecting smaller lumps (3-5cm) perpendicular to the length, from front to rear at each single location. This worked rather well.

Red rectangle is the field. Blue lines are sampling strips.

At home I investigated the soil and it seems to be silty clay.

The large lump is about half a metre!

As a reference I cycled a little bit further to the north and collected samples in a potato-field. This time I used my sampling drill, but only superficial (soil was still very compact) and in the between the potato beds as much as possible. Although the structure of the field looked quite different, the soil turned out to be clay after all.

At home I dried the lumps to the air and crushed them all to smaller pieces (3-5 mm) and powder. The five sample locations were investigated separately. The soil from the potato-field was already fine-grained. The samples were (air) dried for about 2 hours, but not sieved yet.

(After several hours of additional air-drying, samples from the same field were sieved, mixed thoroughly and kept in a bottle. Two days later an additional set of measurements was performed on the samples of the two fields.)

Sample before and after crushing

Taking 40 grams of all the samples, I decided to add only 40 grams of water, obtaining a 1:1 mixture. It was possible to determine both pH and conductivity in the slurry.

After two rounds of measurements, a single drop of a 0.755 M CaCl2 solution was added to the slurries. This could cause the CaCl2 concentration to get up to 1 mM and I was curious to see what would happen to the pH. Then I added 0.5 ml of the same stock-solution, raising the CaCl2 concentration close to 0.01 M (10mM), to see whether the pH would drop with half a point indeed (as stated in the literature quoted previously) .

(I also determined the conductivity to see whether it would be close to the value of the 0.01 M solution, but is was lower than calculated and much more different than the error of less than 10% would explain. It’s not clear why).

To cut a long story short: throughout the field the pH and conductivity values were very, very similar.

The 99% confidentiality interval is for the samples of the “van Wagtendonk” field.

Adding 0.5 ml of 0.755 M CaCl2 solution to the slurry holding 40 grams of water, caused the pH to drop about half a point indeed (range 7.47 – 7.52)! Those values are not shown in the graph.

The 99% confidentiality interval is for the samples of the “van Wagtendonk” field.

The conductivity values presented are for the 1:1 soil water slurry. Of course the values after adding Calcium Chloride are left out, because then the conductivity was increased artificially (to the range 1531 – 1868).

Actually the pH and conductivity values were a surprise, as they were completely different from the ones for the soil samples taken in the wood.

Here all pH values were above 7, meaning they were at the alkaline side rather than acid. I already noticed a lot of shells were in the soil. Not surprising for the bottom of a former sea of course, but the calcium compound might be able to absorb acid. Then the conductivity was much higher than for the samples from the wood.  The 99% confidentiality interval for the population was 239 – 383 The range was mainly this broad because one location was an outlier. Why? I wouldn’t know (I certainly was not eating crisps while working with the samples!). The range is still very different from the low values in the woods (those were twenty times lower, for the sandy soil with humus). The values were also very different from the lakes in Flevoland (three times higher), but that’s comparing apples and oranges of course.

The potato-field, despite the very different macroscopic structure had more or less the same values for pH and conductivity.

Location of the potato field, used as a reference with sampling area (red).

The question is how influential the use of fertilisers may be. For comparison samples can be taken from the recreational wood to the other side of the van Wagtendonkpad. But the next update will be about the samples after additional drying, sieving and combining all samples at field-level. Then we will present another river.


The Merwede, yet another river

Some time ago I wrote that it would be good to get some samples outside in the South, to be sure that we are not looking at local anomalies. Although the Maas would have been better, I was happy to stop at Gorinchem to get some samples from the Merwede – the extension of the Waal, actually a side river of the Rhine. It’s still the Rhine-IJssel system, of course, but not too close to Almere.

It was a sunny afternoon and the temperature was 26oC.

Usually I take three samples and measure them, keeping only the third to check at home and to see what happens to the pH and conductivity over time, but not this time.


To get close to the water I had to walk at the landing stage of the ferry and a jetty nearby. Walking back and forth would take time and it was rather busy at the quay, so I took three samples at once from the landing stage and then three additional ones a bit closer to the river itself. Then I walked back to my car and started measuring, taking advantage of the parking fee I paid.

The results were not surprising at all, because by now we know what to expect!

The samples differed a little bit, but only within a rather narrow bandwidth: 0.2 or 0.3 for the pH and 10-30 µS/cm for the conductivity. The two locations were not too different either, with their averages close to each other, although the standard errors were quite different (much lower for location 2).

Earlier measurements were also done during warm weather and looking at the range of all pooled samples gathered earlier (see: “What if .. all lakes were actually at the same pH?”) we predicted (based on the samples collected back then) that 99% of the future samples would be in the range 7.66 – 8.78 and indeed the six Merwede samples, each measured twice, all fit in this range.  So there is nothing unexpected about the Merwede’s pH.

Conductivity we didn’t have back then, so I took some recent observations, pooled and sorted them and created a graph. The Merwede samples have the lowest conductivity until now.

At home, about four hours later the measurements were repeated (the bottles had been closed at “a summer’s room” temperature and were kept in the dark). The pH went down for most of the individual samples (paired measurements, both duplicate readings), but the effect was weak and smaller than the measuring error. The conductivity however, was really higher for every sample (paired measurements, both duplicate readings). The increase was between 27 and 32 µS/cm.

For both moments the averages of two measurements per sample (1-6) were taken.

Because of the high temperature hardly any additional CO2 will have been dissolved and this will probably explain the rather stable pH. But what could have caused the increased conductivity?


Finally, soil samples: the results

The previous post was about the methodology. This time the results will be presented. Let’s start with some graphs.

Conductivity is in µS/cm

The different approaches of the measurements allowed me to do a lot of calculations and see the differences between them. As expected the 0.01 M CaCl2 solution lowered the pH, but even more than expected – in some cases up to a whole point instead of a half. Of course there are no conductivity values for this case.

Drying and sieving increases the pH a little and reduces the conductivity – roughly between 20% and 30%, although the 21B value looks like an outlier. When measuring the pH in the slurry rather than in the fluid above, the difference is smaller. However, for the conductivity it’s the opposite, because all the sand in the slurry will increase the electrical resistance – hence showing a lower conductivity.

In general these soil samples were rather acidic (somewhere in the range of an orange, pineapple or an apple, but not quite getting to the level of a lime (see this picture for a visualisation of the pH values of fruit), although the mixture with the CaCl2 solution (0.01 M) got below pH 3!

This is unlike what we observed for the water in lakes, with pH values usually above 8 and even over 9.5 in some situations!

Diluting the 2:1 mixture to a 5:1 ratio increases the pH as expected with 0.3 or 0.4. No need to do that, because it was only meant to be able to filter the slurry, but filtering seems to be a bad idea anyway. It’s not just the pH being pushed up, but the impact seems to be very unpredictable (remember that we used rinsed filters). The same goes for the conductivity, becoming unpredictable after filtering. In general the dilution from 1:2 to 1:5 reduces the conductivity, between 25% and 45% as expected. We better stick to the 2:1 approach in the future as it is the best alternative for 1:1.

Drying and sieving is a good idea anyway, because it also allows us to mix several samples from the same area. This will provide a more stable picture of an area, when e.g. ten samples from an area are combined. So next time we will go for dried and sieved samples, being mixed with twice their weight of water. Measuring of the pH will be done in the slurry, but for the conductivity it should be the fluid above.

Perhaps the usage of 0.01 M CaCl2 2:1 will be investigated again, to see whether the impact will become more predictable for larger amounts of samples.

Still one question left: is there a correlation between the pH and the conductivity like we observed before in lake water? The answer is yes.

Although the impact is not too strong, the correlation is clear. Especially when looking within the sample values rather high correlations were observed. Below there are two tables after all. Sorry!

Looking within the samples a clear correlation is observed, but it could be the method!

The dried and sieved samples also show a clear correlation, but for the others it’s rather weak.

In the end the conclusion is that lake and river water and soil samples seem to be opponents. High pH versus low pH and rather high conductivity versus rather low values. Actually I didn’t expect soils samples to have such a low conductivity, although I knew the soil is getting rather acidic in the Netherlands as a result of the ammonia and nitrite/nitrate deposits. In the future I will try and determine those concentrations as well, with my new kit.


Finally, soil samples

After a long range of posts about water samples, I gathered my first set of soil samples in the wood.

This picture is from another moment. To the right is only a partial sample.

Water samples are so much easier than soil samples, because they can be measured right away and then water is mixed well, although it’s not completely homogeneous as we saw earlier.

Now it was time to practice with soil samples and investigate all the different methods used in labs all over the world. Of course I’m still concentrating on pH and conductivity. The analysis of salts will come later.

In the Southern part of the “Spanderswoud in the middle of the Netherlands, I took two samples at two locations each, close to two different posts indicating a walking route (numbers 10 and 21). Four samples in total and although this sounds like duplicate sampling, soil does not really mix, so the results can be very different and actually it is better to mix a large number of samples for one location to avoid incidental anomalies. In this case I wanted to know the similarities and differences, so I kept the four apart and split the samples for different kinds of treatments.

First, I split each sample into halves. One half was left to dry in the air for about two hours and later they were sieved though a rather coarse net (openings about 2 mm). The remaining pieces were discarded. By the way, I used a kitchen scale to obtain the right portions.

Air drying and sieving. Bottom right the remainder is shown

During the drying process the other half was split again (except for on sample, which turned out to be rather small).

The two quarters were treated differently. The first ones were mixed with an equal weight of water, but then I noticed the slurry was too thick to put even an instrument in, so I switched to a ratio of 2:1 adding the same amount of water again. This time it was more fluid, but my attempt to filter it was a bit pathetic, because nothing happened. The filter, although rather porous, was clogged immediately! Fortunately I only tried to filter the first one and still I managed to measure the slurry in the filter once, but most likely it was already compromised by the filter as I discovered later on.

Determining tare, weighing total and adding water 2:1 to obtain a slurry

The 2:1 water to soil sample slurries were measured in two rounds, both for pH and conductivity. I also noticed that the fluid on top had a higher pH value than the slurry at the bottom of the cup. That’s probably why it is advised to put the pH meter in the slurry [page 14]. At the same time the conductivity would go down, probably because a lot of sand is present in the slurry, increasing the electrical resistance! Apart from the two normal pH measurements I also did a single observation pushing the pH meter into the slurry.

The second quarter of the sample was mixed with a 0.01 M CaCl2 solution. In a previous post I told about the pH being completely unchanged and it was still the same as demineralised water: slightly acidic, with a pH of about 5.85. However, mixing it with the soil sample in the same 2:1 ratio, the pH of the mixture turned out to be much lower than for the demineralised water. This is a known effect [the addition of the salt does lower the pH by
about 0.5 pH units compared to soil pH in water (Schofield and Taylor 1955; Courchesne et al. 1995)] Of course the conductivity wasn’t measured now, because the CaCl2 would mask everything.

Then I went on with the dried and sieved samples, mixing them with demineralised water only (2:1 weight/weight again). Again the pH and conductivity were measured in two rounds and then I doubled the weight by adding more (demineralised) water, basically creating a 5:1 mixture.

Two more rounds of measurements followed – again with an additional series of pH measured in the slurry. If no buffering capacity was present in the soil, the addition of more water would increase the pH by 10log(2) or  10log (2.5), depending on how the soil itself will count (as a volume or not – probably in between). This would be an increase in the pH of 0.3 – 0.4.

Measuring pH and conductivity in the slurry

Finally I tried and filtered the 5:1 slurry and took additional measurements for pH and conductivity. Neither filtering nor measuring the filtrate were a great success. Although I used a rinsed filter, the pH still went up. Later I started thinking that it’s probably not some kind of substance in the filter, but a property of the paper ad- or absorbing (H+) ions. For now that’s all about the methodology. Next time I will present the results and my conclusions.


Does filter paper influence pH or conductivity?

It may sound like a weird question, but I’m planning to investigate soil samples and then it will be better to filter the slurry before measuring the pH and conductivity. Of course those values should not be influenced largely by the filter paper. Laboratory filters are rather expensive and then I’m not talking about the glass filters (those will be completely neutral for both pH and conductivity), but about paper filters. A cheap and sufficient alternative is to use coffee filters like the ones sold in supermarkets. When testing the latter I noticed that both conductivity and pH went up when passing demineralised water through it. How would this be for professional filter paper?

For my birthday I got (among several other things) an analysis kit for nitrate, nitrite and ammonia and professional filter paper from the Braumarkt. The analysis kit will be very useful for the next series of samples, but first I focused at the filtering, to be sure the measurements will not be biased.

Coffee filter to the left, Braumarkt filter to the right. Funnel used below.

For this post I used the data of four professional paper filters and four simple coffee filters. Five times (at first ten, but the second series of five is not really informative) I passed a new amount (probably 10 ml) of demineralised water through the filter and tested both pH and conductivity.

In the graphs I present the results for the professional filters (four experiments) and the coffee filters (also four experiments).

Numbers below show the number of times an amount of demineralised water was passed through.
Numbers as indicated above.

The conclusions are simple: the first sample of water passing through the filter will have a much higher conductivity and pH, so it is better to rinse once before use.

The pH change is not very relevant, because the demineralised water is somewhere between 5.5 and 6.00 because of the CO2 solved in it. No minerals are in (that’s the meaning of demineralised water) and therefore there is no buffering capacity. Moving towards a pH of 7 is not impressive, but it should not be much higher.

The conductivity going up – it’s zero for demineralised water – is something else. Somehow an increase in conductivity means that a salt (or acid or base) is being added (non-ionic substances like e.g. sugar, don’t influence the conductivity). Now I had to cut the professional filter paper and touching it with my hands will have added some salt. For reference: dipping my finger in the water for a fraction of a second increased the conductivity to 10 µS/cm immediately. Dipping twice doubled the value. These values correspond roughly with 5 and 10 parts per million of salt respectively (TDS – Total Dissolved Salt). That is a not a lot at all. Five or ten grams of salt in a cubic metre of water and for the 10 ml we are talking about probably 50 or 100 micrograms!

We already observed the conductivity of lake water being around 1000 µS/cm, so the error introduced would be about 1%, but for soil samples it might be even less influential. The funny part is that the professional filter paper is thicker and denser, so the water will remain in contact longer. The thin and porous coffee filters allow the water to pass quickly. Those effects are reflected in the graphs as well. In the end it’s clear the cheap coffee filter will do, but all filters should be rinsed with demineralised water once before using them.

After some initial experiments I wanted to throw away the used filters, but then I realised those are the ones rinsed with demineralised water and I should keep them!


Water samples and pH – adding conductivity

Having this new device, the conductivity meter, it was time to broaden the activities for water samples. Since the lake “Weerwater” is close to my home in Almere, that’s where I started again. Now it is a couple of weeks later and summer came. No reason to expect the pH values to be the same, but the combination with conductivity could be interesting.

By the way, I purchased a book on limnology (“water science”) by Wetzel. It turned out to be a very thorough description of all aspects of water, but especially pH and conductivity are not discussed in depth. Yet some interesting information is in I will use.

To be able to compare the pH measurements with the original samples, two months ago, I went to four of the same locations: the long jetty near Schippersplein, the entry of the small harbour at Stedenwijk, the boat-like jetty at the Maastrichtkwartier in Stedenwijk and the end of a strip of land close to the “Phantasy beach”. Two days later I added the environment of the pumping station “Blocq van Cuffeler” and the “IJmeer, near Marinastrand Poort” again. You may remember that for the latter two I had some issues with the use of my pH meter, so the corrected values are shown here. The new pH values and the conductivity are added this for the most recent measurements (averages shown)

2 months agoJuly
Jetty near Schipperpl.8,308,34914
Harbour Stedenw.8,028,51876
Boat Maastrichtkw.8,168,46914
Strip near Phantasy beach8,298,48929
Blocq van Cuffeler (“bridge”)8,648,271074
Ijmeerdijk, near Godendreef8,698,47774

Again three different samples were measured and as always I kept the third sample for reference at home and for checking developments in time.

With only a small number of samples we cannot draw conclusions, but it seems like the pH and the conductivity are correlated in a negative way. All individual samples were combined for day one, but also for two days later. We already know the pH goes down in time (with samples kept in closed bottles, at room temperature in the dark), but the conductivity is going up at the same time. The effect will be weaker than the trend-line suggests, because of some outliers. Yet, if only the averages for day 1 are used the effect is still present.Funny enough the correlation with a small number of values (table above) is -0.73, but with all sample-values for day 1 and 3 it is only -0.48 Despite the limited number of values, both correlation values differ significantly from 0 (t-test).

The pH in Dutch lakes will be mainly related to Calcium and Carbonates. Somehow a lower pH means more solved salts or acids are present (the conductivity is an indication of the Total Dissolved Salts. Most likely the closed bottles in the dark will produce CO2, solving in water and dissociating to H+ and CO22- or HCO2and this will reduce the pH. Some CaCO3 might even solve again by turning into Ca2+ and 2 HCO3 but that’s just speculation.

We can see that adding the conductivity is really useful, because it provides us with more information to put the pieces of the puzzle together. Now the question is whether the extreme pH values, observed earlier, will also show a lower conductivity.


Why I’m a bit silent now: some other Citizen Science

During the last couple of weeks I posted on a regular basis, but now I’ll remain a bit silent. It’s for a good reason and I already told a little bit about it. Firstly I got my original blog site back (and posts had to be copied from the old to the new site) and secondly I was involved in a “Citizen Science” project concerning water quality in the Netherlands.

For this project I promised to visit fine locations and four were in Almere, so that was rather close to where I live. The fifth spot was further away and rather complicated. After all it wasn’t as easy as expected and this weekend it took me about eight hours in total (checking the locations, preparing, sampling, taking pictures, counting and filling out the online form, after a lot of preparation during the weekend before).

Because of this I wasn’t able to take my own samples. Originally the plan was to revisit some locations and look after both pH and conductivity, but this has to wait for another weekend now. The posts are usually about measurements done in the previous week or weekend and this time I’m  running dry because of a lack of time.

I’ll be back and hope to talk about the combination of pH and conductivity, moving on to soil samples as well.


Juggling with brine

No pH this time! I already mentioned the nice conductivity meter I purchased, but to be sure it was nice, I had to investigate the precision and accuracy. Precision is not an issue. It’s just repeatedly measuring salt solutions. Accuracy is different, because some calibration fluid has to be present. It wasn’t my plan to pay a lot for an overpriced bottle of brine and I was thinking about a way to make it myself. Although the kitchen salt I used holds a little bit of Iodine, it’s within the error margin. Only about 2.2 mg per 100 gram and even when the NaI will be more soluble, it won’t exceed the 10 mg in total (on 400 g of total salt)

Since at 26o C 359 grams of kitchen salt can be dissolved in (demineralised) water, I decided to prepare a saturated solution by adding excess salt (400 grams) to the water (obtaining 1 litre in total) and leave it for a couple of days, mixing it now and then. Finally I mixed it again, waited for an hour or so and took 16.3 ml from what I expected to be a 6.14 M solution (M is mole per litre, so 6.14 * 58.4 grams in a litre – rounded that’s 359 gram).  16.3 ml would hold more or less a 0.1 mole and adding water to a total volume of 100 ml would make the result 1 M. From this stock,  I started to dilute 10 ml of solution in 100 ml of water or 5 ml of solution in 50 ml. After a couple of steps I obtained 0.1, 0.2 and 0.3 M and even 0.01, 0.02 and 0.03 M. By the way, for the small volumes I used an old 5 ml pipette (divided in tens of a ml, but 0.03 is possible to read) and for the larger volumes an 100 ml graduated cylinder.

To compensate for errors I took two different pathways and compared the results. The conductivity meter gave very consistent results and the values were on a straight line (more or less, as it is actually slightly bent, especially at higher concentrations).

Be aware that this graph is biased as explained below. The concentrations are 13% lower than shown on the X-axis

Demineralised water had conductivity 0, like expected. The 1, 2 and 3 milli mole per litre were 10, 21 and 31 micro-Siemens/cm. Very precise, but when I prepared a 0.05% solution, with an expected conductivity of 1014 µS/cm the result was only 880 µS/cm. How could I end up with an error of 13%? I was very aware of the errors I introduced with my dilutions and it could go up to 6%, but it was more likely to be less, because those were random errors, annihilating each other most of the time.

Finally I decided to craft a very precise balance. It’s not too hard when the arms are long and a perpendicular is used (relying on gravity). Actually I used to have a real milligram laboratory balance myself in the past, but this would do for once in a decade.

The small piece of tape is for calibration! It’s very sensitive. Note that the weight to the bottom is taking care of the perpendicular.

An old Dutch dime is 1.5 gram, so I obtained this amount of salt, added it to 100 ml of water (the volume of the salt is negligible) and diluted again 10 ml to 100 total volume.

The conductivity of the new solution was… 1010 µS/cm – as predicted from the literature.

Now I was really confused and I wanted to understand why my saturated NaCl solution would be 13% off. Finally I realised that – different from dilutions, always being about adding water to the final volume, solubility is about putting the salt into a litre of water. Doing it like that, the volume of the salt will be added and actually it’s not 359 grams to be solved in one litre of final solution, but 359 grams with a litre of water added. The density of NaCl is 2.170 g/ml and therefore the 359 grams will add a volume of 359/2,17 to the 1000 ml, making it a total of = 1165,5 ml holding those 359 grams.

This means the saturated solution was not 6.14 M, but only 5.27 M – an error of about 14%

Therefore the conductivity of my (supposedly) 0.05% solution (derived from the saturated solution) would not be 1014, but only 870 µS/cm! Now I had been able to prove the accuracy of the conductivity meter in two different ways.

Still being curious about the CaCl2 solution I prepared a couple of posts ago (meant for soil samples to come), I checked the 0.01 M solution and indeed it was 2.4 mS/cm.

Tap water, by the way was about 415 µS/cm – rather rich in minerals, it seems.

For those who want to know more about the simple and inexpensive device doing so well: this is the Amazon-link.


The original scientassist.wordpress.com – it took some time…

For a while no new posts appeared. That’s because I got my original blog address back and had to do some work to get it up and running. The result is much better, because even I wasn’t able to remember the scientassist[large-number-looking-like-a-phone-number].wordpress.com!

However, it was not the only reason. In the Netherlands an NGO called “Natuur & Millieu” (Nature & Environment) was looking for Citizen Scientists to investigate water quality, so I took the training. I am a scientist and a citizen after all and the training only took a couple of hours. The measurements are one-offs and will take probably a day in total (five rather extensive investigations on different locations, sending samples to a lab as well). For the Dutch readers who are interested to join: here’s the URL. If you don’t have time, some money will also do 🙂 One of the more time consuming activities was the construction of a Secchi disk. A simple but elegant device to measure the transparency of the water. Mine is not the size of an old LP record, but a smaller one works well. Nowadays a CD/DVD is common, but I used the disk of an angle grinder.

Secchi disk

Then – and this will be discussed in the next post – I purchased a conductivity meter. A small Chinese device and I didn’t expect much of it, but it turned out to be very precise and accurate!

It was the one with a lot of positive reviews whilst the negative ones were hardly relevant to me (some received a broken or used one – that’s not the manufacturer and some others didn’t quite understand how to use it). It doesn’t look fancy, but after a several hours of measurements, I’m impressed. Next time I will tell you about my adventures with different concentrations of NaCl solved in water (brine, simply put).


More surprises: details along the dyke

Nearly two weeks ago I took samples along the dyke (Oostvaardersdijk) from Lelystad, via Almere to Zeewolde. The post about this set of samples showed a map with pH values. Somewhere at the beginning of the parking IJmeerdijk the pH value was rather high: 8.7 Previously the same high value also came up, so now I was looking for more detail. This time I planned to look at the other side of the bridge as well. We live in the “new land” (polder), but after crossing the bridge the “old land” is reached. Would things be different there?

To be honest, I didn’t think it would be possible to defeat the high pH score of the Hoornse Plas in the province of Groningen (9.57), but this time things were even more extreme. Again it was in an area where people are chilling out at the beach (a very small beach), SUP-ing (Stand-Up Paddle-boarding –  I didn’t know it was an abbreviation), rowing or swimming.

The cause of the high pH is probably not at the Almere side after all, because there the pH values over there were much lower. Here the top-score was 9.70 – soapy water again! I still don’t know if this is caused by human presence.

Yet it was a bit strange that about 750 metres away from the small beach the pH was still 9.63! Because the sampling was done at this point first, I wondered whether it could be the shells at the beach. After mixing a sample of those with some demineralised water, the pH I got was only 8.08. Surely not an explanation.

The good news is that – despite the high pH – a lot of little fish were swimming in all the Muiderberg-locations. At least this would mean the water was not toxic. Most likely the ammonia-concentration was low. Of course the pH will go up when the weather is hot, but now it was only around 25oC

Getting samples was not easy this time and I had to walk at some of the locations, looking for a good spot (not at a beach or harbour itself, but closer to open water). Again the pH-values are plotted on a map.

Again all values are the average of three samples (third one kept) and again I measured the values of the third sample again when back at home – also checking the reference buffers before and after. Standard deviations (of the samples) were 0.015 or less. The reference values were never more than 0.03 off, so the measurements are reliable. Of course the values remeasured at home were about 0.05 – 0.2 lower, as explained in a previous post.

By now it is time to move on and investigate soil as well. Conductivity will also be added in the future and hopefully some other analytical values.


Astounding results!

A couple of posts ago I told you about my trip to the North and the three samples I took. It was only three samples because traffic was much slower than expected and I also had to make a detour. Because of this – although I left two hours early – I could not take the other three samples. It wasn’t a big issue, because nearly all samples until then showed a pH between 7.98 and 8.92. The first samples were at a small recreational lake in the suburbs of the city of Groningen, called “Kardingerplas”. The pH values for samples 1 to 3 were 8.28, 8.32, 8.31, so no surprise with an average of 8.30 and a standard deviation of 0.02. Quite normal. What surprise could the other lakes bring? Well, BIG surprises!

The next set of samples already brought completely unexpected results. To start with, the lake I selected is a bit complex. It has a section called “Hoornseplas” which is meant for swimming. The other part has to names: “Hoornsemeer” (Norhtern part) and Paterswoldsemeer” (Southern part). In the past (until about forty years ago) two lakes were in place, but taking away the barrier between them, just one lake remained still having two names. There is even another lake to the south (Friescheveenplas), but I didn’t know about it until I was finding out what I had been looking at, later when back at home. To provide an overview of the situation, my sampling points are plotted at the map below.

Coming at the “Hoornseplas” I observed a dam and I wondered whether this was not just a (legally) specified section of the “Hoornsemeer” as it looked like this small lake was completely separated from the larger one. To be sure, I sampled both sides of the dam and the pH was so different that there was no doubt: the two parts are not connected! The part called “Hoornseplas” showed pH values 9.56, 9.56 and 9.58 for the three samples taken, with an average of 9.57 and a standard deviation of 0.01. A pH of over 9.5 (!) is soapy water and perhaps that just what it is. If a lot of people are swimming there, bringing their sunscreens, lotions and shampoos, it might become more of a bath tub than a lake. It’s not dangerous to swim in (then a bathtub full of soap water would be as well), but surely plants and animals won’t love it.

Close to the other side of dam, there was a bridge and there the pH turned out to be only 8.17 (8.16, 8.15 and 8.18). To be sure I took a reference sample at a jetty about 10 metres away and it was 8.15, so we can be sure that the dam is a real dam, separating the two parts allowing them to differ 1.4 points in pH. Probably the dam is in place to save the real lake(s) from the devastating influence of the people bathing there. If so, it was a wise decision. The picture below shows the situation clearly – but there’s another surprise to come!

It’s not the end of the surprises. Sampling at another location, in the part called “Paterswoldsemeer”, I got very inconsistent results between my first and second sample. Standing on a jetty I took the first sample to the left and the second to the right. The jetty is hovering over the water and no barrier at all so I didn’t expect a pH of 8.55 to the left side and a pH of 7.95 to the right and thought this would be a mistake of some kind. Instead of taking a third sample only, I took another pair to the left and to the right and the results showed the same gap. Then I noticed the right side was close to a canal. The water coming from there could influence the water to the right of the jetty, leaving the water to the left more or less unchanged. There was a small bridge over the canal and I took a sample from there and indeed the pH there was only 7.70! At the end of the jetty, where the two flows would be mixed a bit more, the pH was 8.11. To the far left, far from the influence of the canal, the pH was 8.66. It is a very difficult story, but shown on a map it will be very clear what is going on there.


Originally I added an animated gif, to show the situation. Unfortunately then twitter wouldn’t show the link to this post. The animated gif had to be replaced by some pictures.

The jetty from where the sampling was done – to the left, to the right and later also at the end.
View to the left
View to the right (end of the canal)
Looking back at the right side, seeing the bridge over the canal.

Samples – the days after

You will remember that every third sample from the “two lakes and the river” was kept. The samples were stored in the dark, but at room temperature. Earlier samples from the Weerwater and several other lakes were kept as well. Later the same was done while “sampling along the dyke”, also keeping every third sample.

For all samples, the pH dropped gradually during the days after and even during the hours after sampling, although the difference is quite small for only a couple of hours (usually only 0.01 or 0.02). Yet it was a reason to measure immediately, at the source location.

The decrease of the pH has to be related to biological activity, e.g. bacteria or algae (although we don’t know for sure). Adding cyanide could stop the process, but would change the pH at the same time. I don’t have cyanide and I don’t want to have it either, so that’s not an option. Heating would stop biological processes, but could also influence the pH and the same goes for cooling, although that could be a valid option. We’ll investigate it in the future.

For now it’s enough to observe this drop again and again. However, the change is not the same for every sample!

The (third) samples were measured two times (for the later sets in two cycles, but the standard deviation is always small: 0.01 – 0.02 and the drop is mostly ten times higher) with a check of the buffer’s pH before and after (usually no need for calibration). The results are presented in the graphs below, with some space for the days without a measurement.

Be aware that the scale of the Y-axis is slightly different from previous graphs.

Several lakes

The “Weerwater” is repeated for comparison (see previous post on this subject). This time the graph got the same Y-scale as the others, empty positions were added for days without measurement to make the time-scale more realistic and the demineralised water reference was removed.

The rate of change is not predictable. For one sample it can be 0.2 and for another one 0.6 or even more! Sometimes the pH seems to go up again, but I think that’s because of homogeneity or the sample. I later experiments the sample was stirred while measuring and then the results are more stable.

What really strikes me is that when I was obtaining the Weerwater samples and the days after, the temperature was really high: about 27oC The other samples were taken in a period with a temperature of about 10oC lower, around 17oC. Yet the change of the pH doesn’t seem to be very different. I wanted to show the pH-change against the original pH in a graph. Some approaches were too complex, but with a trick it worked out: a lot of values were available for day 5 and the ones missing had day 4 and day 6 values. I interpolated the latter (taking the average) and used them as day 5 values to get a complete graph. The result is clear: the more the initial pH differs from neutrality, the faster it is moving towards neutrality (although hardly even going beyond 8 within this period).


Sampling along the dyke

Previously the pH of the Markermeer turned out to be really high. Even when putting the pH-values of all primary samples together, those values were really high in comparison with the other lakes and the river IJssel. Of course I could take some additional samples at the same location, but why not sample along the whole dyke (in Almere)? Starting in the Northern part where the municipality of Lelystad begins and then sampling every couple of kilometres (if possible) until we hit Zeewolde.

Parking Kernplaats

This time I went by car, because it would be a trip of over 50 km (it turned out to be 75), and the wind was really strong. Apart from this now I know that taking and measuring three samples, marking the bottles and taking a picture of the environment takes about a quarter of an hour, so ten samples would add two and a half hours to the journey. The disadvantage is that a car can only be parked at specific parking spots, but the day before sampling I checked all the nice locations on Google maps and sent myself a list of addresses.

The parking places are often at viewpoints, having their own names (in Dutch) like “ Nonnetje”,  “Kuifeend” (both names of birds) or more prosaic names like “Pampushaven” (just meaning Harbour of Pampus). Below I show a map of all the locations where I actually took my three samples, with locations in degrees.

Again, I discarded sample 0 as it was only for rinsing the bottle, then I measured samples one to three and kept the third. Values were written down in a notebook, together with the time. A picture of the location was taken for reference and the bottles were marked.

When I came home, a second measurement was done, but only to exclude potential errors (third sample only of course). All values turned out to be very close to the original one for sample 3 – mostly about 0.04 points lower as expected (based on previous investigations, showing a gradual pH drop over time). The high values were moving a bit more towards a lower pH. Of course the earliest samples were older (about four hours) than the later ones (less than an hour), but the pH-shift was not really different. Before measuring again, the pH meter was checked (not calibrated) with two buffer values (6.86 and 9.18). The deviation was still only a couple of hundreds and afterwards (measuring in two rounds) the values were still very close, so no doubt about the accuracy of the pH meter. I present the pH values with two decimals, but keep in mind that the precision is not: the standard deviation will take care of it.

However, one was an exception, because the pH went up, but this was a special sample, taken closely to the Blocq van Cuffeler, the huge pumping station. The three samples had the lowest values measured in the whole series, but were very different from each other, suggesting the water was not very homogeneous. That’s not a surprise, being close to a pumping station bringing (rain-) water from the canals into the lake. Then the third sample was also the one with the lowest original value of those three samples.

Here we will stick to the average of the three values obtained directly after sampling.

The averages and standard deviations (of the population) were calculated and the averages were put on the map. For those who care about significance the averages and standard deviations (for significance testing keep in mind that dividing by the square root 3 provides the standard error of the mean) will be available in a table below. Here we present the pH values in the map, because it offers the best visualisation.

There is no doubt that the pH gets up when we get closer to parking IJmeerdijk near Almere Poort (a very nice parking by the way, where a lot of people are chilling out). Despite this local high, it’s very clear now that the pH will never be below 8.00 in this area (and probably this applies to all Dutch lakes and rivers, let’s see). For the IJmeerdijk value of 8.70 (again!), the question is whether some kind of dump is lowering the pH locally. It could also be the opposite, this being the spot where hardly any fresh water comes in. We could see the effect of fresh water close to the pumping (pH relatively low) station and around it’s outlet to the lake.

Below pictures of the environment of the sampling points are presented. Be aware that the actual sampling point may be a hundred metres arpart. The real sampling point is just a bunch of basalt rocks, like the picture at “Kuifeend”. I get as close to the water as possible (sometimes really dangerous because of slippery algae and unstable rocks) and I use a stick with a cord to be able to sample as far from the rocks as possible. Then I will avoid beaches, harbours and closed areas as much as possible, staying close to the open water. The sampling point “near da Vincipad is an exception and so is the small bay close to the pumping staion “Blocq van Cuffeler”. At the map it is al very clear, especially when using the coordinates.


Blocq van Cuffeler (outside dam)
Blocq van Cuffeler (inside bay)
Pampushaven near da Vincipad
IJmeerdijk, parking
Near Musweg

What if … all lakes were actually at the same pH?

This may seem a peculiar statement, because the samples expressed very different pH values. And yet… let’s imagine all the water is just not completely homogeneous, with the differences being just very local. Taking three samples in a row we already noticed the differences. The pH meter also has a random error of probably 0.03 (between different measurements – in a calibration buffer the values are quite stable) and all this together could lead to (small) fluctuations of the observed values.

To test this idea (it’s not a real hypothesis, but more of a thought experiment), we could take all measurements on fresh samples (ignoring the secondary values for the days after) and put them all together. It doesn’t matter whether it was for different samples (triplo) or repeated measurements of the same sample (duplo) and it’s even not relevant which time of the day we took the samples.

In total we got 72 primary values and we won’t use the correction for the drift of the meter (observed during the first measurements) right now.

We can calculated a 99% Confidence Interval (99% CI) for the population (all samples to be taken,  using a Z-value of 2,576, so minus and plus 2,576 times the standard deviation of the sampled values,  to obtain the lower and upper boundary of the 99% CI). We can expect 99% of our future samples to be within this bandwidth (but beware, this will only apply if the distribution is Gaussian).

Dividing the previous standard deviation by the square root of the number of samples minus 1 (that would be sqrt(71) = 8.43), we can calculate a Confidence interval for the mean.

If we would calculate the mean (taking the average in Excel) for a reasonable set of future samples, we would get a value between those boundaries – at least with a certainty of 99%, so once in a while it could be outside, but that should be very rare.

Determining the median: sorting all values, we can take the value between #36 and #37, but both are 8.16 so that’s easy.  Obviously the median is lower than the mean, so the distribution is skewed, we have outliers or both. The mean is very sensitive to outliers, but the median is not. The sorted values can also be visualised in a graph.

72 samples ordered by pH

Then we can see clearly the outliers to the right. Those are the values we corrected, because the values measured for the calibration buffers were too high. Another way to visualise the distribution is the creation of classes.

Now we can see that the shape is not a nice Gaussian bell-curve at all. Even not when the outliers are taken out. But… the bias could be caused by my selection of course. The Weerwater has many more values than e.g. the Markermeer!

What we should do is sample those high pH lakes (especially Markermeer and IJmeer) again and see whether they are really more alkaline* than expected.

To be honest, taking them out won’t change a lot and by now it seems that the pH is nearly always 8 or more. On the other hand, as already mentioned, there is a strong connection with the Ijssel, and because the Ijssel branches from the Rhine, we should consider the South or the East of the Netherlands as well. Let’s see where we get.

* Alkaline (and alkalinity) is now often defined as “resistance of the pH to acid”, effectively being buffering power. However, a long time ago when I was doing the research for my Master’s Degree in biochemistry, we used the word as the opposite of acidity. The word “basicity” to indicate a high pH was not used at all.


An adventure with Calcium Chloride

The last couple of weeks were all about water samples, but the plan is to sample soil as well. That’s why I bought a nice instrument, allowing me to access several depths, without digging. It’s a kind of hollow drill, called a “soil sample probe”.

Actually I still have to practice, but something else was also important. Soil is not a solution, so how do we measure the pH?  Adding water is an option indeed, but does it affect the amount of acid or base? I read a lot about the different options. Some take fresh samples and mix them with water. Others wait until the soil is air dry. The amount of water differs as well between the labs. Some take an equal mass of water to mix with the soil sample, which seems quite reasonable. Others use five or ten times the volume.

Then I encountered several articles talking about Calcium Chloride (I don’t mention the formula yet, because that’s the issue for this post). Adding it would change the pH, but provide more reliable values. The article used says: “Soil pH measured in water is the pH closest to the pH of soil solution in the field (this is true for soils with low electrical conductivity and for soils that are not fertilized), but is dependent on the degree of dilution (the soil to solution ratio). Measuring soil pH in a matrix of 0.01 M CaCl 2 ,as opposed to water, has certain advantages, but the addition of the salt does lower the pH by about 0.5 pH units compared to soil pH in water (Schofield and Taylor 1955; Courchesne et al. 1995).“.

This means I should give it a try, so I decided to get some. Of course it is possible to purchase expensive, high quality chemicals, but here we are talking about mixing it with soil. Calcium Chloride is well known as a de-humidifier  and that’s the cheap way to lay hands on it. The advice was to make a stock solution of 1 M and dilute it to 0,01 M (10 mM) and so I did.

The substance is hygroscopic, so I wanted to dry it well before weighing the right amount – 111 grams per mole according to my own calculation – confirmed by a couple of websites to be sure. After 20 minutes in the oven at 75 degrees Celcius or so, created my stock solution (111 grams of CaCl2 filled with demineralised water to 1 liter in total, to get 1M – or at least that’s what I thought)  and took 10 ml of it, diluting it to 1 liter to get the 0.01 M.

The article (it’s actually more like a manual) mentioned the conductivity bandwidth (between 2.24 and 2:40 mS cm-1). It seemed a bit superfluous, but I gave it a try. Physics is fun as well, isn’t it? The setting was rather primitive, but sufficient for the purpose. I took about half a meter of and electric PVC pipe and put a plug at one side, pierced with metal screw. The other side I put on a clamp and connected my Ohm meter. First I used a digital one, but I still get confused and measured again using my old analogue one. After some calculations (the diameter of the pipe was 13.5 mm, so the resistance had to be multiplied by 1.43 and divided by 0.45 (length of the tube). Then the resistance was converted to conductivity, by taking the reciprocal. It was about 25% too low (1.7 mS cm-1)!

After some thinking and reading I realised that my pearls – whether heated in the oven or not – were more like CaCl2.2H2O. Otherwise it would have been a powder rather than pearls.

Adding 2 times a water molecule makes a mole:  111+36 = 147 gram. This meant my stock solution was actually 0.755 M and the diluted 0.01 M was actually only 7.55 mM – lacking 25% of the salt. The latter was corrected easily by adding some additional stock solution, but the next time I should take 13.25 ml of stock solution and fill it to 1 litre, to get the 10 mM after all.

I will also investigate if tap-water will do, because in the end it’s cheaper and saves a lot of shopping (I already used ten litres of demineralised water).

Next time we will get back to water samples again.


Two more lakes (and a river)

By now nearly all the samples showed a pH of 8 or higher. Yet I went on measuring water samples. Going to Zwolle, I planned three more sample locations. Two were still more or less connected: the Velumemeer and the Ijssel. The latter is a river, feeding the Ijsselmeer, but indirectly also the Markermeer, IJmeer, Gooimeer and Veluwemeer, so it could be that in the end we are more or less looking at the pH of the Ijssel’s water.

The third location however, was less connected, because it was a lake in Zwolle’s neighbourhood “Stadhagen”. The lake’s name is “Milligenplas”.

The three sample points are marked in the Google Maps screenshot of a part of the journey

Like the full day sampling I took three samples at slightly different spots (a meter or several metres apart). After rinsing with the sampled water, the first two samples were discarded, but the third was kept for later investigations. As a result the direct measurement was in triplo.

The results won’t be sursprising by now, although the rather high pH of the IJssel is still remarkable!

The bandwidth of the graph is rather narrow this time: 7.9 – 8.5

Next time we will have a look at all primary values we obtained until now.


Measuring during a full day

One of the possibilities explaining the high pH would be photosynthetic activity, removing the CO2 from the water. The samples are taking close to the surface, but the Weerwater is not a deep lake. At the same time the wind will stir the water and additional CO2 will be solved, so it’s hard to tell which effect is stronger. The only thing we know is that photosynthesis needs light and therefore it makes sense to measure the pH in different light conditions. To investigate the effect I decided to take samples every couple of hours, at just one point of the Weerwater, close to where I live.

The Maastrichtkwartier has a funny area with a jetty, together shaped – more or less – as a boat.

Because the location is close to where I live, I was able to walk down to this spot, do some sampling and return to home again. Earlier I noticed that the pH of the samples will vary with the (micro-) location. Even samples taken just one metre apart (some in a sunny place, others more in the shadow) will differ in their pH value. That’s why I took three different samples this time, only about a couple of metres apart. The samples were measured immediately after bringing them up and discarded afterwards, except for the third one. The third one was measured again 46-64 hours later, to check whether the pH changed a lot.

I have to admit that I sampled around midnight and early in the morning again, but I was not enough of a fanatic to get up in the middle of the night as well. At least I was not willing to do so for now, not having seen any effects yet.

Again the buffers were measured before and after the experiment and the accuracy is ok, with the precision being within a couple of hundreds.

Averages of three samples for every moment during the day, with an 99% Confidentiality Interval

Because I measured three samples, I was able to determine the standard deviation and calculate a 99% Confidence Interval (green lines). The whole Y-axis only covers 1 full pH point, so it is clear that the chance of the water actually being neutral is negligible. At the same time the differences between evening or midnight and sunny conditions during the day are also very small. It is interesting that for some moments the standard deviation was larger than for others. A very small part of it is the caused by the pH meter, but the location seems to be more influential.

In the end, the high pH found in several lakes can’t be the result of plants taking away CO2. There is a small effect, probably causing the pH to change with tens of a point, but certainly not bringing the pH up a whole point from neutrality (from 7 to 8 or more)!

In scientific terms our hypotheses would have been:

H0 photosynthesis does not change the pH from neutrality to a value around 8 (night values won’t be neutral, that is).

H1 photosynthesis is responsible for the high pH found in the lakes (and during the night the pH will drop to neutral or even acid values).

Not being able to reject the H0 is not proof that H1 is false, but in this case the variations are so small that H1 is nowhere in reach! In the past I plead for a “confirmation interval” in addition to the “rejection interval” (outside of the 95% or 99% confidentiality interval), to avoid wrong conclusions and here things are rather clear. There might be a very small photosynthetic effect, but it’s not enough to explain a pH of 8 or even higher. We could even say that the pH did not change significantly during the day and we have to look for another explanation.

46-64 hours later (depending on the age of the sample) I measured all third samples again and put them in a graph. The original values of the third samples (not the averages of the three shown in the graph above, because only sample 3 was kept!) are shown as a grey reference-line. Unlike previous series, the pH drop was rather small, but the temperature was also 10 oC lower. If biological activity is causing the drop, then it makes sense that the change was much smaller now, because of the lower temperature.

The same scale was used in both graphs, so it’s easier to compare the fluctuations during the day and the shift after two or three days.


Broadening the area

Of course Almere is only a part of the Netherlands. Going on a trip to the North I selected some other sample points, which I would visit during my trip. The Ijsselmeer was sampled at two points: Lelystad and Lemmer. It’s separated from the Markermeer, so I wondered if the pH would be very different. Then in Frisia the Tjeukemeer was sampled at the parking “De Lanen” (actually under the A6). Traffic was not as good as expected and I had to ignore the other sample points planned for, but for now it will do.

Let’s see what the results were. It would be a rather long day and the measurement of the samples at home would be ten hours later. For the other lakes we already noticed that the pH drops a little bit during the days after, so I decided to take my pH meter with me and do the first measurement right after sampling.

By now it won’t be a surprise that the IJsselmeer is also was in the higher pH range, but the same applied to the Tjeukemeer! Back at home I measured the samples again and two days after another time (of course measurements were done in duplo, taking all precautions). The results are shown below.

The calibration was not an issue anymore. The buffers provided very stable values during the last week or so. I don’t put in the graph, because it might be boring. Of course all the data are available in a spreadsheet and I keep my original log papers.

Having collected enough samples to know that al least not all surface water in the Netherlands is acidic, I went looking for explanations. Ammonia seemed to be a possibility, but it’s not very likely, because the water quality is sufficient. Moving to soil samples in the future, it will be interesting to determine the inorganic Nitrogen (NO2, NO3 and NH4+) and I read a lot about all kinds of methods (hoping to find a less complex and less dangerous method than the classic Berthelot in dozens of scholarly articles). Limestone, concrete and all kinds of waste water could explain local elevations of the pH, but not in several different lakes at the same time.

The I found an interesting site saying “An additional cause of elevated pH is high photosynthetic activity, which removes carbon dioxide from water and raises pH”. This would mean that the pH differs throughout the day. Biological activity is also a good explanation for the pH-drop of the samples, because I keep them in the dark, but not cool.

The next investigation might be (very) local again, now looking at the factor time!


More lakes, more surprises!

After the surprise of the Weerwater being rather alkaline, I set up a map of new sample points for different lakes. Biking together with my eldest son, it took 50 km and a couple of hours to collect them all, but let me show you the map before we go on.

The Weerwater was in again, but only with two sample points now. It was clear that the pH was not very different at the sample locations last time, so it made sense to reduce the number of sample locations. The other large lake (Noorderplassen) got two sample locations as well. The water around Almere is formally divided in three lakes with different names, but all these are remainders of the former sea (Zuiderzee), which became a sweet-water lake after closing the dam (Afsluitdijk) 90 years ago. The Markermeer and IJmeer are not separated by any construction, but got different names. The Gooimeer, to the South is connected to the IJmeer, but the connection is rather narrow.

Now the question is, will these lakes be more acidic or at least less alkaline than the Weerwater?

The same (clean) sample bottles were used and rinsed with the lake-water first before collecting the actual sample. Because of the long trip, the measurement was done a couple of hours after sampling, but again we will check what happens to the pH during the days after sampling. The weather remained nice. About 25 oC and still no rain!

The results were completely different from what I expected – again. More alkaline samples and this time even more extreme! The Markermeer scored 8.92, but I still had this issue with the pH meter being slightly off. That’s why I applied a correction formula and presented the corrected values. For the Markermeer the corrected value became 8.69. Still a very high pH. Although I had some trouble getting my pH meter right (fortunately the next time everything became stable), this value is confirmed by an article (although a couple of years old). It says that the pH of the Markermeer is around 8.7.

The good news about the measurements is that the days after I finally managed to get my pH meter stable (rather accurate and precise, using the calibration buffers). After this new attempt the measurements could vary a couple of hundreds, but the average was stable and close to the buffer’s value. I calibrated once and then the values remained stable. The 4.00 buffer was checked less frequently because the samples were in the region above 7.

Seeing the stable results, I was confident to measure the change of the pH of the samples over time again. Like the previous set the samples were kept in a closed bottle, at room temperature and in the dark. During a couple of days after sampling they were measured and stored again.

For most samples the pH will drop slowly, meaning the sample is getting less alkaline. Most likely this is the result of some biological processes. Either the alkaline factor is processed, or some acid is formed (neutralising the sample) or both. Since I don’t have chemical details (yet), we can only guess. By the way: all data are available in a spreadsheet and the original logs will be kept. Please reach out at anrep3d@gmail.com if you are interested.

More lakes to come!


What happened to the pH of the samples during the week after sampling?

In the previous post the conclusion was that the water of our lake (Weerwater) is definitely alkaline. But will the pH change during the days after sampling? A little bit of air was available in the bottle, but it remained sealed, meaning no additional oxygen was present. Measuring during the days after was not planned, but turned out to be a good exercise. I had to know the pH meter and its behaviour, so I measured again the next day, the day after and so on. It turned out that the pH moves towards neutral, but very slowly. Most likely this is the result of biological processes, performed by micro-organisms, like bacteria. Either the alkaline factor is being removed, or neutralising acid is being produced – or both. I don’t see a relationship with the amount of organic matter at the sample location. Especially the “Fantasie beach” (actually I sampled at the end of a jetty) was rather clear, but it is the area where a lot of water skiing is going on!

The first couple of times I was calibrating before the measurement and checked the pH of the buffers afterwards. When I realised there is a small deviation rather soon (minutes), which doesn’t change in the days after I didn’t calibrate the next time, but checked whether the values for the buffers were unchanged. They were.

Next time I will tell about the water samples taken from different lakes in Almere. The old hypothesis of slightly acidic water was revitalised. I expected the “Weerwater” to be anomaly


The pH of our lake is not as expected!

One has to start somewhere and since we live very close to a lake that’s what I preferred.

This lake is the central lake of Almere, called “Weerwater”, meaning “Water again”. It actually describes the history of the lake, because at first the former Zuyderzee (litterally Southern Sea) was closed off by a dam and eventually became a sweet water lake. Then about half of it was drained and became new land (polder) and on this land the town of Almere was built. To get enough sand for the construction works, a huge pit was dug to serve as a sand quarry. After completion of the construction works the pit gradually filled itself with water and was left as a lake, so it was water again after the draining. Now this hole in the sea bottom is the lake “Weerwater”. The bottom of the – rather shallow – lake is covered with water plants and those have to be mown now and then to allow the boats to cross the lake without destroying their propellers. It is about three times the size of the reservoir in Central Park New York.

Of course a sample at a single location would not do, so I selected several points around the lake to take my samples and went on a ride with my bike. Currently the tour is about 12 km. May 7, 2022 was a nice warm day with a temperature of about 25 oC

Sample points lake “Weerwater” Almere, with GPS coordinates

After collecting the samples, I started the pH measurements. Of course I had a hypothesis and it was that areas with more decaying organic material would be slightly more acidic than others. Because of the CO2 in the air I expected the water to be slightly acidic anyway, although not the 5.75 measured in my demineralised water after being shaking for a couple of seconds.

You already know I bought this nice pH meter and had to learn to work with it. At first I calibrated it with the three buffer fluids meant for this purpose, but it seemed like the meter had a “drift” (systematic error), because afterwards the buffer values seemed to be different (after every calibration). Measuring during a couple of days I noticed that the drift only applied to the values shortly after calibration. Then the deviation remains stable, at least for several days. The electrode was rinsed thoroughly after every measurement, especially when buffers where involved because they are capable of changing the pH of non-buffered fluids strongly.

Bottles used to store water samples

The bottles were plastic ones (150 ml), originally holding fruit shots, but thoroughly cleansed with hot and cold tap water and detergent rinsed several times with tap water and demineralised water, which after a while always seems to have a pH below 6 (slightly acidic). As a young boy, over fifty years ago now, I noticed that an inspector taking samples to determine milk quality at a farm, wrote with a pencil on an opaque area on the glass. Now I made a similar opaque area at the bottle’s surface using sand paper. No labels needed!

During sampling at the lake, the bottles were rinsed with lake water once, before taking the sample. Then the bottle was closed and kept at room temperature. The first series of measurements were done only a couple of hours after the collection of the first sample, but additional measurements were performed the days after. The additional hypothesis was that samples with more (microscopic – the fluids were all clear) organic material would tend to get a lower pH (more acidic).

To make a long story short: my original hypotheses about acidity went out of the window. The water was rather alkaline! I’m sure about that, because in between I measured demineralised water and after the measurements I determined the pH of the buffers again (showing more or less the same systematic error again and again). Although the pH meter is sold with a precision of 0.01, I think we have to accept an error of about 0.2 for the accuracy, which is all right. Especially in the alkaline area there is a slight tendency to exaggerate the alkalinity, meaning the measured pH is about 0.2 higher than what it should be according to the calibration buffer. Below the averages of the first three measurements (performed an hour after taking the last sample) are shown (the names identify the location – Dutch names).

Boat Maastrichtkwartier.8,16
Harbour Stedenwijk.8,02
Pier near Schipperplein8,30
Lumièrepark beginning8,27
(point inaccessible)
Fantasie beach8,29

Average pH of three measurements for all Weerwater samples

Below there is an explanation about the systematic error, but the conclusion is that we should take off 0.2 from the measured values to have a more reliable outcome. However, the water will remain alkaline anyway! The values around the lake differ slightly, but in general they are around 8!

Now the question is: how did the water get a pH above 7 (alkaline) instead of the expected value below 7 (acidic)? My new hypothesis was that this could be the result of the construction of the Floriade – the large world exposition for horticulture – or other constructions with concrete going on. That would mean other lakes in Almere would have a lower pH. I will tell you about those measurements another time, after explaining what happened with the pH during the days after collection.

For those who want more detail about the precision and accuracy of the measurements: Learning to know the new pH meter, I calibrated several times and noted down what the measured values of the buffers were after a while. It looked like there was a bit of a drift, but measuring after several days without calibrating, it turns out it’s a rather stable systematic error. To show this systematic error of the buffer measurements, I present the average and standard deviation below. Although there is a 95% confidentiality interval, the deviation from the buffer’s real value is much more interesting. I’ll spare you the details, but we have to correct the measured values by 0.2 towards neutral, reducing the alkalinity. The buffer 4 value is not very stable, but that’s because the focus is not at this range, making it less reliable. The electrode also needs some time to switch several pH points and during the first measurements and I probably didn’t wait long enough in the acidic region. After all the pH meter is very precise and stable, but not completely accurate at the moment. It may be better to wait some time before pressing the calibration button. Still learning!

Buffer 6.86Buffer 4.00Buffer 9.18
Measured (average)6,754,089,36
Standard dev.0,080,390,21
Lower 95% Conf.Int.6,603,298,95
Upper 95% Conf.Int.6,914,869,77

Precise values, but not completely accurate as there is a systematic error


Scientassist is back!

Before and after my graduation I had written a book and done some freelance work, but it was not enough to make a living. So I decided to take a regular job, but nobody was looking for a biochemist! That’s why  I started my additional education in business administration/finance and I was able to get a job at the large hospital, where I already did the research for my Master’s degree.  This time it was about answering business questions and creating reports for the board, with the help of datasets and one of the first personal computers. Gradually I was drawn into the world of business data and processes. Eventually I became a consultant specialised in business process management and (financial) reporting.

The original Scientassist logo!

Yet I wanted to do something more creative, more entrepreneurial. That’s why I founded my company Scientassist, about thirty-three years ago. Despite the description (“bureau for scientific assistance”), I was selling my first product: a map in which areas were coloured automatically, based on an input-file and a set of boundaries. Other products followed, but that’s all water under the bridge. Those adventures can still be read at my old blogs for VRBI and AnRep3D (and even EnRep3D).

Now the question is what this Scientassist blog will be about. Let me explain. Throughout Europe we have a serious problem with Nitrogen deposits and acidification of our soil and water. We know a large part of it is related to our current agricultural practices. We need farmers, but we should help them to change their business models. I was asking myself how I could contribute, going back to my roots. It all starts with providing some insight in the concentrations of Nitrogen-compounds in soil, water and air.

Nitrate pollution linked to agriculture, says new study

Using IoT (Internet of Things) to set up a network of measurement points would be nice. I thought by now we would have cheap but accurate sensors to measure NH3 of NH4+ and NOx, but I was mistaken. Setting up a small chemical lab, like I had in my youth and even after my graduation (actually until I became a father, around the time Scientassist was founded), would not be the best choice either.

Finally I discovered that pretty good pH meters, with a glass electrode, are very affordable now! Since the indirect effect of the Nitrogen-deposits is acidification, I could work with water and soil samples, to create a pH-map of the Netherlands as a first step.

This isn’t sponsored content!

Next time I will tell about what happened from this moment on. Despite my education as a scientist, this blog should be considered as “citizen science” rather than formal scientific research, backed by a university.