Tools for Quantitative Archaeology |
DSPLIT: Wilson and Ward Radiocarbon Analysis Program
Compares and combines radiocarbon samples using the procedure published in Archaeometry by Wilson and Ward (1981; Ward and Wilson 1978).
RUNNING THE PROGRAM
To run the type DSPLIT<Enter>.
The program reads parameters from the keyboard and C14 dating information from a separate data file which may be on any disk drive, provides information on program progress on the screen, and produces an output listing which may either be saved to a file or printed directly.
PROGRAM PROMPTS
File Name for Output Listing {.LST} ?
Name of the file or device for program results. If you want the output to go straight to the printer type PRN; if you want it to go to a file, give a file (or path) name.
Analysis: [M]LE, Case [1], 2[a], 2[b] ?
Wilson and Ward case number M = mle - solve simultaneous equations for theta and sigma for the data set. 1 = case i - determine outliers, only counting error used A = iia - cluster into groups, use counting error and calibration and sunspot error. B = iib - cluster into groups using case iia error and taking sampling error into account.
File Name of Input Data Set {DSPLIT2A.DAT} ?
Enter the name of the file (including drive and directory if necessary) from which the data should be read. The format of the file is described below. To enter data from the keyboard, reply CON to this prompt. The default file name for data input is given in {}, will be given if ENTER is hit. If you reply with a ?, the program will ask for a directory (hit ENTER for the default directory), list all files for the directory, and re-prompt for a file name.
Calibration and Sunspot Errors in Input Data {N} ?
This prompt will be issued only for case i in which calibration and sunspot errors are not used in the calculations. Thus, these error terms are not required in the data file. However, if the data file includes the calibration and sunspot errors, reply Y to this prompt so that the program knows they are present in the input and can then ignore them.
Minimum MLR for Split - from Chi2 table (95%=3.84) {3.84} ?
The minimum MLR to split the sample. This cutoff is derived from a Chi Square distribution. 3.84 is a 95% confidence level, 0.00 allows all splits. Hitting ENTER is the equivalent of replying with the default, enclosed in {}.
Iteration Step to Modify Sigma² Estimate
Use 0.0 for computed step; -25.0 as published
Initial Step {0.0} ?
Iteration step size is requested only for case iib or mle. This number is added at the initial estimation step to arrive at an improved estimate of theta. If a constant other than 0.0 is entered, the subsequent step sizes are found by multiplying by -0.01. If 0.0 is entered the initial step is -10% of the variance of the means and subsequent steps are found by multiplying by -0.1. Entering -25.0 duplicates the published algorithm, but may require a very large number of iterations.
Number of Signif. digits after decimal {10} ?
This parameter is requested for the mle case only. This number is used to determine the accuracy that the program attempts to achieve in estimating theta and sigma. This parameter is limited by the computer's precision. This parameter is limited to 19.
DATA FOR EACH OBSERVATION READ BY THE PROGRAM
The sample data are read in free format from an input file. Except for the label, the sample data are read as integers and are read in free format (separated by blanks, tabs, or commas). Although the program does not require it, each new set of sample data should start a new line. Any string enclosed by #'s (or from a # to the end of the line) is ignored. Using this # convention, identifying comments can be included anywhere in a data file. The sample data for each cases must be entered as follows.
label = a label (maximum length 32) is read for each case. Labels may not include embedded blanks, commas, #'s, or tabs.
age = the age in years bp, e.g. 4710
agevar = the counting error (standard deviation) provided by the lab (as a +/- on the age).
f = the calibration error (standard deviation). This value is not read for case i (but may be included). W&W suggest 50 for ages younger than 2700 bp and 60 for older samples.
g = the sunspot effect error (standard deviation). This value is not read for case i (but may be included). W&W suggest 70 when the effect is suspected, otherwise 0.
SAMPLE DATA SET (DSPLIT2A.TST)
#Ward and Wilson Test Data For Case iia M-912 4401 125 60 0 C-288 4402 141 60 0 M-911 4471 150 60 0 Y-1280 4490 80 60 0 Y-1279 4500 80 60 0 M-195 4525 200 60 0 C-367 5383 250 60 0
SAMPLE DSPLIT RUN
Dsplit: Wilson and Ward Radiocarbon Analysis 5/2/1986 - 11:20: 8 File Name for Output Listing {.LST} ? dsplit2a.lst Analysis: [M]LE, Case [1], 2[a], 2[b] ? A File Name of Input Data Set {DSPLIT2A.DAT} ? dsplit2a.tst Reading Data 7 Number of Observations = 7 Minimum MLR for Split - from Chi2 table (95%=3.84) {3.84} ? Sorting Data Printing Sorted Data Performing Analysis End of Analysis: 0.3 Seconds Analyze Another Dataset {N} ? N
SAMPLE OUTPUT OF DSPLIT RUN
Dsplit: Wilson and Ward Radiocarbon Analysis -5/2/1986 - 11:20: 8 Analysis Case iia Number of Observations = 7 Test Statistic Cutoff = 3.8400 Initial Iteration Step =4.038616E-39 Listing of Sorted Data No Label Age S 1 M-912 4401.0000 138.6542 2 C-288 4402.0000 153.2351 3 M-911 4471.0000 161.5549 4 Y-1280 4490.0000 100.0000 5 Y-1279 4500.0000 100.0000 6 M-195 4525.0000 208.8061 7 C-367 5383.0000 257.0992 Analysis of Wilson and Ward Case iia Split between points 6:M-195 & 7:C-367 - MLR[6:M-195]= 12.1103 Age[6:M-195] = 4525.00 >> M-195 Age[7:C-367] = 5383.00 >> C-367 Ap[1,7]= 4506.73 T[1,7]= 12.7542 Ap[1,6]= 4469.48 T[1,6]= 0.6439 Ap[7,7]= 5383.00 T[7,7]= 0.0000 End of Analysis: 0.3 Seconds
Home | Top | Overview | Ordering | Documentation |
Page Last Updated - 23-Nov-2009