Protocol for AMS radiocarbon dating of plant macrofossil material
Becky Briant and Ian Lawson
1. Get someone to give you the money
- AMS radiocarbon dating, like conventional radiocarbon dating analysis,
is expensive so you will need a source of funding to pay for the analyses.
- Applicants in Britain may apply to the NERC Radiocarbon Committee ( www.gla.ac.uk/nercrcl/ ) or Geochron Laboratories (less money available - www.geochronlabs.com ).
2. Preparation of samples
- Store your bulk samples or cores refrigerated and well-wrapped against fungal attack. However, do think about radiocarbon dating as early as possible, because even well-wrapped cores may go mouldy.
- The first thing to bear in mind before preparation is contamination. If your samples are old (i.e. >25,000 years old), then you have to be very careful, especially since AMS samples are so small. Similarly, if they are younger, the level of precision you require will be higher. Either way, any contamination that you introduce could be crucial. Of course there is also the possibility of contamination that was introduced during burial, but you can't do much about that - though see later. You can do something about this.
- Get yourself the following: a clean lab coat, a hair net (yes, I know, but get it!), powderless gloves, a clean room or laminar flow cabinet if these are available (if not, all the labs in Geography have filtered air input, so use these and try and keep surfaces free from dust by wiping daily with a damp cloth), copious amounts of distilled water, clean forceps/tweezers, a paint brush (preferably not made from animal hair). Wear all the stuff all the time. Think laterally, and be careful.
- General points: You want to make sure that you are working as quickly as possible, so that samples are not hanging around half-processed for long periods of time. Don't get slack - cleanliness is very important, for example even how you wash your glassware - see Vogel reference. Make very detailed notes of what you are doing in a notebook which you take great care of. You will need this when filling in sample submission sheets. Also, be very careful about labelling - don't get depths confused, for example.
- Do not let any alcohol or glycerine or formalin or any other chemical containing carbon, or paper which may give off fibres such as blue roll anywhere near your sample - ever. A full list of contaminants to avoid is on p.8 of Polach and Golson (1966).
- If you have a bulk sample, then the first step is to weigh out 1kg of material (making sure you always keep archive material - at least enough for pollen and LOI). Put this into a bowl with some distilled water to soak. Then set up your (very clean - that includes no detergent residues) stack of sieves (300mm diameter, 250μm, 500μm, 1mm meshes) in a sink with a sediment trap and a running source of distilled water. Wash the material through the sieve stack until the sieve surfaces are full. Decant into clean (preferably brand new) glass jars for the relevant size fractions. Continue until all sediment has been washed through.
- If you have a core, remove at least 5mm around the edge of the core before sampling as this area is where most contaminants reside. Also avoid all samples within 10cm of either end of the core segment. Start by taking a narrow slice of sediment, and increase it if necessary - thin stratigraphic samples = higher precision dates. Make sure you keep archive material. If you are storing your core subsamples before processing, wrap them in tinfoil rather than plastic, as plastic may shed small fibres. When sieving, you may want to use smaller diameter sieves (e.g. 100mm).
- Store residues in the fridge.
- Thoroughly clean your sieves.
- Before you start, you might like to know which plants are dateable. Firstly choose plants whose habitat requirements are consistent with the palaeoenvironment as ascertained from the full floral assemblage (so that reworking is unlikely). Also, terrestrial plants are preferred over aquatic plants, as they photosynthesise directly with the atmosphere and are therefore more likely to have had 14C/12C ratios in equilibrium with it at time of death. This circumvents the problem of reservoir effects in lakes, and of 'hard-water error' in limestone areas. Therefore avoid Potomageton even though they are nice and chunky! However, Carex and Cladium mariscus are dateable (although aquatic taxa) because they photosynthesise above water.
- OK, Now you are ready to start picking. Unless you're very lucky, each sample will be a mystery to you in terms of floral composition. Therefore you'll just have to be patient, and sit there over your microscope in the wet sediment lab until you've picked out stuff that looks like seeds and leaves from your whole sample.
- Store everything in the fridge when you're not working on it, and cover when you go for lunch/coffee etc.
- Picking out plant macrofossils is done wet, and if you can do this using forceps without trashing every seed you try to pick up, then well done. Otherwise, you'll have to use a brush and risk contamination (that's why it's preferable to use a non-organic brush). Alternatively, you could try using a pasteur pipette with a teat on the end. This has been found to be quite effective for charcoal and sometimes for smaller seeds.
- Also pick out molluscs and store these dry and separately in sample tubes.
- Do not pick out beetles, as Russell Coope (if he's willing to look at your beetles) prefers to float them out of residues from which plant macrofossils have been removed.
- Therefore, keep your residues for future reference. (Lump particle sizes together at this stage).
- Once you think you've got all the seeds/leaves that you're going to find, you need to identify them. This is so that you can date sensible things - see above).
- Identification is a learnt skill, so you'll have to ask someone to teach you how to use the reference collection (try Charles Turner (O.U., & associate member of QPG), Mike Field (Coventry), Alan Clapham (Archaeology), or Becky Briant (QPG)).
- However, the first start is to sort your seeds by what looks similar to each other. Then look at reference books in the main microscope lab (Beijerinck: Zadenatlas der nederlandsche flora and Bertch: Fruchte und Samen: Handbucher der praktischen Vorgeschichtsforschung). Then check your seeds against the relevant part of the reference collection, and get someone else to check it too. If you are comparing seeds on a separate surface, use only glass-fibre filter paper to do so.
- You need about 15 milligrams (mg) dry weight of plant material for a date. Do not be deceived by the many optimistic papers published. This is actually quite a lot of material - and the more you have, the more confidence you will be able to place in your date at the end. Bear in mind that what the lab will be interested in is the amount of carbon they can get out of the other end of the prep process. Therefore, if you have a sample where the seeds look weathered, you will probably need more material for a decent date. Typically, organic material is about 40% carbon.
- Once you have sorted and identified your seeds, now comes the moment of truth. There are two advantages with drying your sample. The first is that you then know how much material you have. The second is to do with minimising your contamination. Although you will have been storing everything in the fridge, there may still have been a chance for fungi to grow (according to Wohlfarth et al. 1998 - JQS 13(2) 137-145).
- When drying, place sample into a labelled package of foil and fold over ends, but not so far that air cannot escape as the sample dries. Place all foil packages overnight in a drying oven at 50oC.
- Also sterilise the glass vials (and plastic caps) that you will be placing the samples into at 105oC in a drying oven. This will involve rinsing the tubes and caps with distilled water, placing in an evaporating dish, and covering with another evaporating dish, as below:
- Using a 3 decimal place balance (third d.p. is mg) weigh the samples (tare tubes first). They can then be stored at room temperature.
- When sending samples off, package them like fragile crystal - each sample is worth £400, and Ian did have a situation where one of his vials cracked even inside three layers of bubble wrap.
- Discuss carefully with your laboratory which pretreatment they are going to use on your samples. Olssen (1986) is useful here. If wood samples are submitted, ideally, the cellulose will be extracted. If the sample is too small, then an acid-alkali-acid pretreatment will remove all soluble humic acids, which may be contaminating it with young carbon. If you have plant macrofossils from peat, you should insist on this pretreatment for them as well. If you have plant macrofossils from any other sediment, Olssen (1986) suggests that this should be applied anyway, to remove any cause for doubt. If your laboratory does not think your samples are large enough for this, they will probably only use a pretreatment of dilute HCl. This is OK, but press for acid-alkali-acid if at all possible.
3. Checking the dates when you get them back from the radiocarbon lab
There are several things that you might want to check for. Fortunately for you, there's a spreadsheet that I (Becky) have written which does these things. It's called 'Radiocarbon contamination checks'. You can download this spreadsheet by clicking on the link below;
- If you have old dates (>~30,000 years) you might wonder whether they are actually statistically distinguishable from 'infinite' dates. Radiocarbon ages are quoted with one standard deviation only, which means that there is only 68% likelihood that the true age falls within these limits. 2 s.d. gives 95% certainty, whereas 3 s.d. gives 99% certainty. Therefore, if the range comes close to 45,000 years at 2 s.d., the date might be too close to infinite to tell the difference (Bowman, 1990, p.37).
- The first box in the spreadsheet does just this. Input your sample names, ages and 1 s.d. in the spaces in red, and the ranges at 1 and 2 s.d. are automatically calculated. Then you can decide if you think that the upper limit is close to 45,000 years.
- If you were expecting your dates to be infinite, you might also want to check what amount of contamination by modern carbon would give an age such as you have got. N.B. This only calculates shifts from infinite, not shifts from older ages.
- This is done by assuming that the full amount of activity in the sample (expressed as percentage modern) is due to contamination. This gives a percentage of contamination. The actual weight is given by multiplying this by the actual weight of carbon in the sample (which your lab should be able to tell you).
- If you input the percentage modern activity and weight of carbon into red parts of the second box on the spreadsheet (sample names are copied across automatically), this will calculate the weight of contamination needed to give that age from an infinite sample.
- Great - but how can we tell if this is a meaningful amount of contamination? By eyeballing other samples, I can see that 0.05mg of modern contamination, laboratory-introduced, would be large enough that the probability of it not being detected is very low. Conversely, an infinitely small amount of contamination has a 100% probability of not being detected. Fitting an exponential function to this gives an equation for the probability of contamination not being detected at each weight:
- A calculation in the spreadsheet calculates the probability of such contamination not being detected for you, and then you can classify the samples and low, medium and high likelihood of contamination not being detected, using the guidelines in the spreadsheet.
- Fuller details may be found in Polach and Golson (1966). However, basically you want to know whether the dates overlap. The first way that you do this is by comparing the 2 s.d. ranges.
- Go into the third box on the spreadsheet and paste in your sample details in the groupings that you want to test for similarity (name, age, 1 s.d.). The 2 s.d. ranges will be calculated for you. It is then possible to test by eye whether these overlap and to note which of the samples in the group they overlap with.
- It is also possible to do a more rigorous statistical test of difference (Polach and Golson, 1966). For each group, work out which samples need to be cross-compared to get a full range of comparisons in the group. Then, for each sample, calculate the arithmetic difference and combined s.d. (Combined s.d. is the square root of the sum of squares of the two s.d.s.) The formulae for these are already present in the spreadsheet - you just need to change the cells to which they refer.
- It is then possible to assess whether these ages are significantly different. If the arithmetic difference is less than 2 times the combined s.d., the samples are not significantly different (confidence is especially high when the difference is less than 1 times the s.d.). If the difference is between 2-3 times the combined s.d., the samples are probably significantly different. If the difference is greater than 3 times the combined s.d, it is highly probably that the samples are significantly different. This can be assessed by eye and the outcome recorded in the final column.
4. Useful references
Bowman, S. 1990. Radiocarbon Dating. British Museum Press, London.
Mook, W.G. & Waterbolk, H.T. 1985. Radiocarbon Dating. European Science Foundation Handbooks for Archaeologists No. 3, European Science Foundation, Strasbourg.
Olssen, I.U. 1986. Radiometric Dating. In: Berglund, B.E. (ed.) Handbook of Holocene Palaeoecology and Palaeohydrology. Wiley, Chichester.
Polach, H.A. & Golson, J. 1966. Collection of specimens for radiocarbon dating and interpretation of results. Australian Institute of Aboriginal Studies Manual No.2, Australian National University, Canberra.
Vogel. Analytical Chemistry. In the chemistry laboratory in the Main Labs.