QUB | Archaeology and Palaeoecology | The 14Chrono Centre

Manual for psimpoll and pscomb

psimpoll and pscomb: release notes

• Macintosh OS 9 and earlier: the amount of memory by default allocated to psimpoll is inadequate, and, without remedy, it will likely fail to run completely. Highlight the psimpoll executable, then select 'Get info' from the pulldown 'File' menu. Choose 'Memory', and increase both preferred size and minimum size to some large number (say 1000kb in the first instance). Repeat with a larger number as necessary. And do the same for pscomb.
• Mac OS X: the psimpoll and pscomb files provided are binary files that should be run directly from the Terminal application. Place the binaries in some place such as /usr/local/bin. Then, within Terminal, change directory to the place where you have your psimpoll/pscomb input files. Run the programs by typing their names. A certain amount of unix knowledge and experience would be useful, but pretty much all other file manipulation (including the creation of input files and display / printing of output can be done with more familiar Mac tools. Mac OS X is a new and interesting type of operating system: I would be very interested to hear of comments and experience with it.

psimpoll and pscomb: questions and answers

1. Generating a line diagram after using histograms
2. Importing from Tilia and accessing sums
3. Input file formatting
4. Crashes
5. Can't find input file (Windows)

1. Generating a line diagram after using histograms

   Question: The data set works fine for the production of a histogram (using code md for line 4), but I would prefer it as a line diagram. Everytime I try I get the following error:
   getbuff buffer is overfull. Check that individual input items do not succeed 169 characters.
   Are the 169 characters including spaces and the title? If they were over 169 characters, surely the histogram would not work either?

   Answer: The problem is that line diagram mode doesn't expect to have the histogram widths: just a value for the sample depths. So, if you had the widths (and I assume you did for code 'md'), delete them.
   The buffer is an internal mechanism that doesn't relate to anything particular in the file. It will only fill if there is a problem in the data, so that items get out of order, which is probably the case here. After running psimpoll, you should notice a file called psimpoll.LOG, which will have a copy of the buffer if the program has failed because it is full. This can be a useful way of tracking what went wrong. The typical problem involves seeing that this has multiple values in it, when there should only ever be one (be that a number or a title), and arises when the program is expecting text (a title or taxon name) and getting numbers instead.

2. Importing from Tilia and accessing sums

   Question: I have been trying to get a Tilia 2.0 file to read into psimpoll in order to calculate confidence intervals. However, it reads ok, but won't calculate the confidence intervals. It keeps saying "cannot find sum so no confidence intervals calculated". (This is one of the main reasons I am using the package).
   The file is a standard Tilia 2.0 % file with taxa, lycopodium and sums. from the psimpoll comments it seems to think the sums are just another taxa.
   I enclose the Tilia 2.0 file I have been using and the file in the general format that I have used to input into psimpoll.

   Answer: Unfortunately, psimpoll cannot figure out itself where or what the sum is in the Tilia General Format file. But, with a little help it can be told!
   Run psimpoll on XXXTEST. Read the data in as '%' by depth 'd'. Select menu J, then 'a', to print dataset to a file. Back to main menu, select 'X' to run the program. You should have a PostScript file with the diagram, and everything plotted, including numbers of tablets and the like.
   Now take a look at the dataset you just created, in XXXTDAT (by default). This is a regular psimpoll input file, and can be manipulated as text only (use WordPad on Windows, something equivalent on other systems). You will see the taxa as rows, and these can be marked to define how psimpoll sees them. 'Taxa' that should be ignored (like the number of tablets) can be marked with a '-' as the first character of the name. Most important in this context, mark the Total pollen sum with a '*' as first character. psimpoll then treats this as a sum, uses it for confidence intervals, and plots it as numbers instead of a graph. The bottom of the file should now look like this:

   Dryopteris filix-mas, 0.0000 0.0100 0.0310 0.0070 0.0050 0.0060 ...
   -Exotic (Lycopodium) count, 459.0000 564.0000 1100.0000 1002.0000 ...
   -lycopodium in tablet, 12542.0000 12542.0000 ...
   -no of tablets added, 2.0000 2.0000 2.0000 2.0000 2.0000 2.0000 ...
   -Volume of sediment, 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 ...
   Trees, 0.0770 0.0960 0.1330 0.1430 0.1300 ...
   Shrubs, 0.1060 0.2120 0.3560 0.1680 0.2390 0.5890 ...
   Heaths, 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0020 ...
   Herbs, 0.1830 0.1990 0.1260 0.2230 0.4440 0.2480 0.1930 0.1240 ...
   Aquatics, 0.0070 0.0130 0.0580 0.1130 0.8680 0.5860 0.4250 ...
   Spores, 0.6340 0.4930 0.3850 0.4650 0.1870 0.0370 0.0450 0.0580 ...
   *Total Land Pollen and Spores, 574.0000 604.0000 483.0000 ...
   

   Save the file, run psimpoll on it, this time with confidence interval options, and all should be well.

   The handling of input from Tilia is much improved in psimpoll 4.00 compared with 3.10. If you are using 3.10, and there is a problem, upgrade.

3. Input file formatting

   Question: I have now finally managed to get the textfile transferred to psimpoll. Unfortunately, I get an error message that says "getbuf: buffer is overfull". I have checked the textfile several times, but can't find that any ind. entry exceeds 159 char. What am I doing wrong: could you please help? I enclose the text file as attachment, so you can check.

   Title
   67
   53
   %m
   Depth   5.70    5.71    5.72    5.73    5.74    5.76    5.78    5.79    5.80
   5.81    5.82    5.84    5.86    5.90    5.94    6.04    6.08    6.10    6.14
   6.18    6.22    6.24    6.26    6.30    6.39    6.42    6.46    6.54    6.58
   6.66    6.74    6.82    6.90    6.98    7.06    7.14    7.22    7.30    7.46
   7.54    7.62    7.70    7.78    7.86    7.94    8.02    8.10    8.18    8.32
   8.78    9.16    9.32    9.44
   Juniper         3,5     2,5     2,2     3,0     2,4     4,4     2,8     4,5
   5,5     7,8     4,4     5,0     2,7     3,8     5,8     5,7     8,3     4,2
   11,4    9,3     9,5     2,3     3,0     3,7     4,3     0,5     2,1     1,6
   2,4     2,6     1,5     1,6     2,4     1,3     2,2     1,6     1,4     1,5
   4,1     2,0     1,8     1,6     2,1     1,3     1,8     1,4     1,8     1,9
   1,5     1,6     1,1     0,6     0,8
   Picea   7,1     5,8     8,9     8,2     6,4     8,0     7,3     7,5     7,5
   7,5     6,8     7,2     7,7     4,6     5,4     6,3     3,8     5,5     3,1
   5,4     5,4     7,3     6,1     6,7     7,7     7,3     4,9     2,5     0,9
   0,5     1,0     0,9     0,3     0,3     0,6     0,5     0,6     0,3     0,3
   0,4     0,3     0,2     0,6     0,4     0,1     1,0     0,2     0,5     0,3
   0,0     0,4     0,0     0,3
   

   etc.

   Answer: Four problems with the file:
   1) Omit the word 'depth': just the set of numbers for the depth values
   2) All the taxa names must be followed by a comma 'Juniperus, 0.5 ...' (or be on a line of their own)
   3) Your data uses the European comma instead of the anglo decimal point. Arguably psimpoll should be able to handle this, but unfortunately it doesn't.
   4) The data should be proportions, not percentages (0.5, not 50). Idiosyncratic, for historical reasons, etc. This shouldn't stop it running, but the output will be odd.

   See also the answer to question 1 for more on 'getbuf'.

4. Crashes

   Question: Will psimpoll cause my computer to crash?

   Answer: The short answer is 'no': much effort has gone into ensuring that this will not happen. In general, if there is a problem, psimpoll will exit gracefully, and leave you with a still-functioning operating system. However, there are some exceptions:

   • Memory: psimpoll has been designed to be portable, and there is no portable way of determining how much memory is available on the computer where psimpoll is now working. This could be done by writing separate code for each system, but I haven't had the time (or inclination) to do that. So, if your dataset is too large, psimpoll will run out of space, and a crash may result. This is increasingly uncommon as a problem, as modern computers have more and more RAM available. It is certainly a problem on DOS systems (and this is one of the reasons why there is no psimpoll 4.00 for DOS). Apart from DOS, your dataset will have to be very large indeed before you see any problem (but see note about Macintosh computers, at the top of the page).
   • Input errors: these can lead psimpoll to think it has more (or less) of some variable than it actually does, and allocate the wrong amount of memory. This can lead to crashes. I am attempting to trap this kind of problem, but it is not easy to anticipate. Make sure, especially, that the numbers of zones / dates is correctly specified in input files.
5. Can't find input file (Windows)

   Question: No dates are appearing on my pollen diagram, but I can't find anything wrong with the C14 file. Can you help?

   Answer: The problem is that psimpoll doesn't know that the file has the .txt extension: it assumes a file called xxxxC14 unless told otherwise. Windows has the nasty and disreputable habit of sticking a .txt extension on to text files without telling you (or even showing you) that it has done it. You can fix this by either taking off the .txt extension, or by going into menu E and changing the default file name (and then saving the configuration).

Archaeology and Palaeoecology | 42 Fitzwilliam St | Belfast BT9 6AX | Northern Ireland | tel +44 28 90 97 5136