Polhemus 3Draw Protocol

Leslie F. Marcus
Department of Biology, City University of New York

Having been lucky enough to get a small grant from City University of New York last year to buy a Polhemus 3Draw, I collected a lot of data on small bovids this summer. Skulls - 102 landmarks, jaws about 19 (data on portable computer), and metacarpals - 17 landmarks.

I worked out a procedure that I rather like and will describe here - all of the components are off the shelf, except for a small program I wrote in Quick Basic that anyone is welcome to.

I. Setting up the instrument - with Santiago Reig's help.

Well aware of metal interference with the device, and problems with cables crossing etc. we devised the following.

Since the stylus cable is long enough - we passed the cable from the box over some hook, or something in the ceiling - leaving enough free space to move the stylus around. That way all of the cables are fixed once and for all. Also you won't drop the stylus anymore. A little Velcro on the stylus and the side of the board helps also to keep it in place. (lost the point twice - once in clay; and once in the groove around the board. Found it fortunately - some redesign is needed there!) Couldn't stand the on-stylus button, and Polhemus' foot pedal was too stiff. Use a $4.00 Radio Shack model (shielded cable and all) and place in other hand. Will design a plastic switch. When the switch is near part of the board - it can throw the measurements off a lot. I like a table that you can walk all around - used a stand for sculpting about 15" x 15" first time, with board big enough to hold the device. Now on top of a narrow wooden book case. Keeps you on your toes. Also designed (with help of Mammalogy in AMNH) a jig for holding the skull or other objects. A solid piece of lucite about 5 x 9 inches and 3/4" thick - heavy, with 1/2" holes drilled. A couple of uprights - one to go in foramen magnum with some clay; other with a lucite V to hold the front of the small skulls we were doing. Could reach all points - top and bottom without moving the instrument and quite solid. Works on human skulls as well. Metacarpals fit nicely in the V at mid-shaft with a little clay, so that when the specimen was mounted I could see all landmarks easily. We did mark the landmarks with an erasable pencil - as that made the landmarks easier to see, when moving around the tower.

II. Programming Procedure

Decided to use LOTUS 123 to take in the data. To do that I needed an interface program that would recognize the ASCII string from the RS232 on the Polhemus and pass it to the computer RS232, and then put it where I want it. The Software Wedge (available from Fowler among others) does just that. It takes in a string of characters, and then divides up the string into x,y,z and puts the three pieces wherever you want. You have to learn how to do the string dissection - but all done in one morning. This is very versatile program - and there is a DOS version ($169) and a windows version ($199).

So here is how I do it. I have configured the Software Wedge to read in x,y,z and put the three coordinates in three consecutive cells in a row, and then move to the first cell of the next row - waiting for me to enter the next data item.

I wanted metric data (you wouldn't know it from above), and I didn't want to deal with the orientation data - so I also wrote a little program to tweak the 3Draw to convert the inches to mm., and put it in point mode.

All of this was put together with a .bat file that 1. Ran a compiled few line Quick Basic program to access the 3Draw, convert inches to mm., and drop the last three values. 2. Run the software wedge with the parsing procedure I had developed. 3. Load Lotus 123 and point to the first cell ready to take measurements.

III. Why Lotus and its use.

Lotus has been around. I use 2.1 because I am old -fashioned (Quattro, Excel, newer Lotus etc., DB anything would work as well; or could take data directly into a statistics package). I have grown to respect Lotus' simplicity and ease of use; and convenient output.

I make a template of the first row and column using the Titles feature in Lotus, which write protects them, and leaves the words when you keep going down or across the spread sheet. The first column holds my landmark, names; and the first row x,y,z. I use a couple of rows to put in specimen specific data as well.

At first we measured everything twice - so I used 6 columns. Then in the 7th column of data (eighth cell) I put a little Macro to give instant feedback of the Euclidean distance between the 1st set of measurements and the second using the Pythagorean theorem. I felt that less than 1 mm. difference was tolerable.

I like lotus because if I get off - wrong row or column, I can reset the cursor on the computer and repeat or go over my errors. I don't look at the monitor much - but check myself from time to time. Errors happen from time to time, and correcting the procedure is very versatile. Only use the keyboard when I have a new specimen (or backup etc.). I also replicate the template rows for additional specimens (never more than 10 to a file) quickly using the LOTUS copy feature. My actual template then consists of names of landmarks etc. for 10 specimens. We are finding that for an older computer data entry slows down if you try to make the spread sheet too large. I save all files as LOTUS .wk. files.

IV. Output from Lotus and Further Analysis

If you output just the specimen data and coordinates as Unformatted to a file, you get an ASCII file that is easy to process elsewhere. GRF-ND - I like to visualize my data using GRF-ND. If I set up the spread sheet right, the data is ready to go to GRF-ND - though some of the specimen specific data may get in the way. But you can easily see if you have a sensible set of measurements, and they look like the specimen if you have decent coverage. The design of your link file in GRF-ND is important here.

My preferred way is to take the data into SAS, and then put output new files from SAS for GRF-ND, and/or NTSYS. SAS also finds where there are any errors - in the original Lotus *.prn file, and it is fairly fast to find which specimen caused the problem, and fix the data - either in the ASCII file, or back in LOTUS and re-save it. I also compute a number of traditional distances and do some bivariate scatter grams on these in SAS. After I have the result file from GRF-ND - I like to use INSIGHT to make 3D plots of individual GLS aligned landmarks.

V. Some other points.

I used a 486 Notebook and worked like a charm. Did not have color, so not so nice in GRF- ND etc.; but adequate. The computer went down near the end of the data collecting trip, and I was fortunate didn't happen earlier. In the future I plan to take a powerful hand-held that has the capability of doing all of my steps through Lotus for just taking data, so that I am not left without data collection capability.

I went to another museum and was able to set up and take data in less than an hour. By the way I use a hanging type soft suit-bag as a carrying case (about $40) for the 3Draw and always carry it with me. It always fits in the overhead on 10 flights (a variety of planes), and the staff never bothered me (I liked the Air Bus best for space). One fit was a little tight, and closed the door a little harder than I would like. The electronic box fit well in the side pocket. Over the shoulder strap makes it easy to carry around, and all cables and power supply fit in bag as well.


The information provided above was last updated Oct. 12, 1995 by Les Marcus. For current information, contact Dennis Slice () or post a message to the morphmet listserver ( ).
File updated February 13, 2009.