Getting Organised for Useful Analytical Results

1 IMPORTANCE OF COMMUNICATION AND PLANNING

In the scenario described below, poor communication between an analyst and a researcher leads the analyst to do work on the samples that does not give what the researcher wanted and expected. Both the researcher and the analyst need essential information from each other to ensure that the work done on the samples is suitable, useful and well received by the researcher; and satisfying for the analyst.

While reading the scenario you are asked to consider the situation that the analyst has to address. At the end of it there are some explanatory notes but, before turning to them, you are asked to think about how the unsatisfactory outcome might have been avoided.

Suppose that the following conversation (which unfortunately is not entirely fanciful) took place between a researcher and an analyst in the laboratory:

Researcher: I am going to harvest some plants from an experiment next week. I need to know the amount of P in them. Can you do that analysis for me?

Analyst: Yes, P determination in plant material is one of our standard procedures – when do you want to bring them in?

Researcher: Well, if I harvest them on Thursday, can I bring them first thing Friday morning?

Analyst: I'll check my work diary ... yes that will be all right. That's agreed then, I'll see you on Friday morning, with the samples.

However, there was a slight change of plan. The researcher had to leave early, but he entrusted the samples to a colleague to take them to the laboratory. So on Friday morning the samples duly arrived in five carrier bags with a note to the analyst saying:

'Here are the plant samples for analysis of P. I hope the results will be ready when I return in 2 weeks time'.

The carrier bags were labelled P0, P1, P2, P3 and P4, and each bag contained four young cotton plants, complete with roots and with soil adhering (Figure 1.1).

Perhaps this analyst has not received samples from this researcher before and the researcher has not had an analysis done previously at this laboratory. So the analyst has to use her initiative in deciding how to deal with these samples.

The samples cannot be kept until the researcher returns:

• Do you foresee any difficulties for the analyst in making decisions on how to proceed?

• What would you advise her to do?

• What material should she take for analysis:

– All the material provided?

– Plants with soil removed?

– Plants with soil and roots removed?

– Leaves only?

• How many samples are there:

– Five samples, one per treatment – each made by bulking together material from four plants?

– 20 samples – i.e. four replicates of one plant from each treatment?

• Do you think these decisions are important in relation to the purpose of the analysis or for any other reasons?

The work was at risk because the analyst and researcher each assumed too much of the other – at this stage they had not communicated adequately. Probably the researcher did not know enough about the process of analysis to question properly the service being offered. Probably the analyst is used to doing a particular procedure and assumed that it would match what was expected or needed in this case.

The analyst could not confirm any points with the researcher, but, being a sensible and helpful person, she proceeded on what seemed to her to be reasonable assumptions:

• Cut off the roots and soil from the plants.

• Take five aluminium trays, label them P0 to P4 and put the four plant tops from each bag into the appropriate tray.

• Put the trays in the oven to dry the samples at 80 °C overnight.

• Crush the dried plants in each tray and grind up all the material in a tray together to pass through a 1 mm sieve to make a homogeneous sample from each tray.

• Continue with the determination of the concentration of P (g kg-1 in the five samples.

When the researcher returned, the laboratory was pleased to be able to report the results they had obtained (Table 1.1).

Now did the researcher have some comments and questions:

• It looks as though the foliar spray treatment of P4 was very effective.

• Are the results statistically significant?

• There was about 1 kg of soil in each pot, so does the g kg-1 mean the same as g P taken up by the plant per pot?

• I want to work out what proportion of the P added was in the plants in each treatment.

What would you expect the analyst's responses to be?

• We didn't wash the plants so some of the P determined in the sprayed ones may be only dried P on the surface, but not absorbed in the plants.

(Which probably means: you didn't tell me the plants were sprayed.)

• We can't do a test of statistical significance because there is no replication.

(Which probably means: you should have labelled the plants as separate samples showing that there were four replicates of each treatment.)

• The g kg-1 means g P per 1000 g of dry plant material. (Which probably means: I thought anyone would know that.)

• We don't know how much dry material there was per pot so we can't calculate how much P was used per pot.

(Which probably means: why didn't you weigh the plants or ask us to; or at least measure them in some way.)

• Here is our invoice for £50 + VAT.

(Which probably means: well it's your loss, not mine.)

So both were very dissatisfied with the results of their efforts and no doubt each felt that it was all the fault of the other.

1.1 Activity

Take a few minutes to think about this situation. Is there any other information that the analyst should have and any other information that the researcher might think it is necessary to provide? List on a separate piece of paper as many points as you can that the analyst or the researcher need to know about the work requested:

• What the researcher needs to know.

• What the analyst needs to know.

1.2 Can We Avoid Misunderstandings?

Quite often a researcher has not thought through all that he or she is asking the laboratory to do and so is unprepared to provide all this relevant information. They may even feel that the analyst is being obstructive in demanding it. But you can see how things may go badly wrong if points are not clarified at an early stage.

Together the analyst and the researcher can work out a strategy to try to ensure that the analytical results obtained are as useful as possible to the purpose intended. It may be necessary for the researcher to organise transport from the field to the laboratory or to arrange for plot yields to be weighed in the field. Perhaps the samples will need to be kept in a refrigerator, pending their further preparation, so the analyst may need to organise that space is available in the refrigerator when required.

It would be wise of the researcher to write down the points of strategy agreed with the analyst and to leave a copy in the laboratory as confirmation of the arrangements. Such records form part of a workfile in a laboratory accredited to ISO/IEC 17025 (see Chapter 4, Section 6.1).

Questions:

• Why does the analyst want to know what types of plants are to be harvested?

• What problems may arise if the amount of sample material provided is (a) too large or (b) too small?

1.3 Notes on this Section

The researcher needs to know, for example:

• The set charge (if any) per sample for the test.

• Whether there may be a further charge depending on the amount of work required for sample preparation.

• An estimate of the time the analysis will take.

• The form in which the results will be presented.

• Whether, assuming the analysis goes well, the information produced will be useful and likely to add value to the experiment.

• If the amount of material in each sample is adequate or too little or too much.

The analyst needs to know:

• How many samples there will be.

• The deadline (if any) for reporting the results.

• What plants are being harvested.

• What material will be received (i.e. tops, leaves or whole plants).

• What will be the size of the sample (i.e. how may it be sub-sampled: fresh or dried).

• Whether the samples need washing (i.e. whether the plots will have received any sprays or dressings containing P).

• Whether any further work is required on the samples (i.e. are there implications regarding how the samples should be treated or stored?).

• Whether the researcher wants to know the P concentration in the plant material, or some measure of total plant content of P per plot (i.e. will the data collected allow the required calculations to be made?).

• If the weight of whole plants (fresh or dry) is needed what steps are to be taken to remove soil (or growing medium) and will the plants be really fresh when received in the laboratory?

• Overall, will the samples be taken in a satisfactory manner for the work to be done – especially with regard to organisation, labelling and avoidance of contamination?

The questions above raise the following issues:

• The analyst wants to know what plants are to be sampled because:

– She may know or be able to look up the approximate range of contents to be expected.

– She may know or expect some particular requirements in sampling, preparation or analytical method for certain types of plant material.

• Problems that may arise if the amount of sample material is (a) too large or (b) too small are:

– Difficulty in dealing with a large bulk of samples in the laboratory due to limited size of bench space, containers and ovens available.

– The small amount of sample taken for analysis may not be representative unless the whole sample is homogenised, which may be time consuming for large samples.

– Amount of material may be insufficient for all analysis and checks required.

– Small samples may be easily contaminated in the mill or grinder.

2 ONCE THE SAMPLES ARE IN THE LABORATORY

The researcher should send or bring the samples, with a sample list, having checked that the samples delivered match the list. Any discrepancies or missing samples should have been noted on the list, together with any special instructions agreed – e.g. freezing, cleaning, drying, etc. On receipt, the analyst should first check the samples against the list, note any discrepancies, and check that any special instructions are taken in hand.

Suppose you are both the researcher and the analyst. You are not likely to have that initial conversation about the samples with yourself. However, you will still need to think through what you want to achieve from the analysis and how you must deal with the samples to achieve it. You must ensure that the points such as you have listed are addressed, and that the relevant information is being taken into account.

2.1 Keeping Records (DON'T FORGET – that you will not remember)

For useful tips, see Rafferty (1999).

• Have a book, preferably hard backed, in which to write all ideas, plans, discussions with your supervisor and laboratory staff, advice, instructions, problems encountered, sample lists and experimental results.

• Put your name on the outside of the book and your name and address on the inside.

• Make sure that you always date your work.

• Do not keep records on scraps of paper – they will get lost, mislaid, wiped clean in the laundry or used to mop up spills.

3 PROBLEMS WITH SAMPLE MATERIAL

Usually we are concerned to get a sample that is 'representative' of the whole of a bulk material, or a fluid or a tissue. Sampling has to take account of the fact that only a small amount of sample is collected from a relatively vast bulk. Often it is necessary to take a number of random samples throughout the material. Any portion of the material represented by the sample must have an equal chance of being included. If we sample only from the parts that are easy to reach, we may be introducing bias.

On the other hand, we may want to introduce some selective sampling, for example, to avoid some non-representative areas – such as the gateway or headlands in a field. Again, we may want to divide the bulk material into different sampling units – such as from different areas of a warehouse or at different heights in the hold of a ship or in a silo or at different depths in a soil.

In the laboratory, we are more often concerned with taking a test sample that is representative of the laboratory sample received. We need first to create a homogeneous laboratory sample from the material submitted to reduce the variability between different test samples. Depending on the type of material involved, sample preparation, sample storage and sub-sampling may all have some influence on the analytical results eventually obtained. It is essential that the researcher and the analyst both understand exactly what is wanted from the analysis in order to avoid inappropriate and irretrievable decisions at this stage.

Some samples do not remain well mixed when dried and handled in the laboratory. For example, a sample composed of sand and compost will easily separate out into its different components simply on standing for a few weeks, so it needs to be thoroughly mixed each time a test sample is taken. Samples that contain a mixture of coarse and fine particles are similarly difficult to keep homogeneous. A sample made up of whole plants will, when dried, become a heterogeneous mixture of leaves and stems. It requires special care to keep the sample together and mill it down to make a uniform material.

The bulk of a sample received in the laboratory may be too big for convenience of storage or for further preparation. We have to decide how and at what point the sample size can be reduced or sub-sampled. If the entire sample received can be homogenised then subsequent sub-sampling is not a problem. Otherwise special care is required to ensure that the sub-sample taken is again fully representative of the material received.

With very wet samples – such as silage or sediments – there is the potential for the soluble components of interest to be lost during sample preparation. Some components e.g. ammonia or organic compounds may also be volatilised or metabolised during storage and preparation. Such samples may need to be frozen for storage and kept frozen throughout the sample preparation and as test samples are taken.

Microbiological activity may continue in the samples and change the constituents, e.g. changes in nitrate content of soil and extracts unless they are stored at below 4 °C. Samples may be altered if they are dried too severely. Silage material, for example, may show a reduction in N content if it is dried at 100 °C.

By convention, soil analysis is done on 'fine earth'. This is air-dried soil, ground and sieved through a 2 mm screen after removal of stones and roots larger than 2 mm diameter. Unfortunately there is not always an easy distinction between 'stones', 'weathered rock' such as soft chalk and hard lumps of soil. So it may be difficult to standardise the preparation of the laboratory sample.

4 IS THE AMOUNT OF WORK REQUIRED FEASIBLE?

Next we will look at the question of what we can do if the work required would take too long or be beyond the physical or financial resources available.

Suppose you want to determine the changes in nutrient contents of a grass crop from field trials at six locations – each with factorial treatments of two levels (0 and 1) of N, P and K fertilisers, i.e. eight treatments:

N0K0P0 N0K1P0 N0K0P1 N0K1P1
N1K0P0 N1K1P0 N1K0P1 N1K1P1

For the agronomic management aspect of the trial the plots would be cut immediately before the fertiliser treatments are applied and then another 12 times at two-weekly intervals. The agronomists would dry and weigh the whole cut from each plot and then hand the material over for the analytical studies.

For the investigation of plant nutrients, the draft plan is to take from each cut a sample for analysis of N, P, K, Ca, Mg, B, Fe, Mn, Cu and Zn. So there would be over 100 samples per month for six months – April, May, June, July, August and September. You would need to check with the laboratory to see if they can handle that many samples over that period.

To do the full study as proposed, there would be: 8 treatments x 6 locations x 13 sample times x 10 determinations = 6240 determinations.

Assuming that the laboratory can deal with the samples, accepting that their charges are: £2 per determination plus £1 per sample for sample preparation; then this will cost £13104.

But suppose you have only £1200 available in your project for the analysis – decide how you would spend that amount to get as much useful information as you can from the trial. You will have to reappraise the requirement for analysis with a view to cutting down the number of samples or determinations – perhaps by reducing the number of sampling times or bulking some samples together.

• You can save £1 per sample if you can do all the sample preparation work for the laboratory.

• What precautions have to be taken in bulking material from different cuts?

• Will you be able to calculate the total removal of nutrients in the grass cut over the period of the experiment?

Think about the questions raised and the possible options you have before looking at the proposed strategy in the notes below.

4.1 Notes on this Section

Various possible options may lead to alternative solutions but a suggested strategy to work within the resources available is as follows:

• You agree to do all the sample preparation work – saving £1 per sample.

• We have to cut down on the number of samples. While we want to keep sufficient replication, the main objective is to investigate the effects of applied fertilisers on nutrient uptake, rather than a comparison of locations – we can choose to do analysis on three locations, rather than six.

• If we bulk together too many cuts we may miss or dilute any treatment effect, however it is reasonable to analyse the initial cut (before fertilisers) and then to combine the material from each following two consecutive cuts. This will give a total of seven sampling times rather than thirteen. So now we have: 8 treatments x 3 locations x 7 sample times = 168 samples.

• Suppose we do only N, P and K analysis on all samples, this will amount to 5004 determinations; then we could do the remaining analyses on the control plots (N0P0K0) and the complete fertiliser plots (N1P1K1) and only on the initial cut and a composite of the following 12 cuts. This would give an additional: 2 treatments x 3 locations x 2 sample times x 7 elements = 84 determinations.

• This would give a total of 588 determinations, costing £1176.