The National Student Research Center

E-Journal of Student Research: Science

Volume 5, Number 3, January, 1997


The National Student Research Center is dedicated to promoting student research and the use of the scientific method in all subject areas across the curriculum, especially science and math.

For more information contact:

John I. Swang, Ph.D.
Founder/Director
National Student Research Center
2024 Livingston Street
Mandeville, Louisiana 70448
U.S.A.
E-Mail: nsrcmms@communique.net
http://youth.net/nsrc/nsrc.html


TABLE OF CONTENTS

  1. Examining The Factors That Affect Random Events Through Coin Flips And Jingle Cups
  2. Is There A Connection Between The Fractal Dimension Of A Rorschach Inkblot And The Usual Clinical Response?
  3. The Fractal Dimensions of Sponges: Is There A Connection?
  4. The Effect Of Different Formation Methods On Fractal Dimension Of Squish Fractals
  5. Termites: The Fractal Dimensions Of Their Random Paths
  6. Heat Conduction And Its Relation To The Diffusion Of Particles Through A Chamber
  7. Bumble Ball- Random Walker Or Not?
  8. Comparison Of The Growth Patterns In Forest And Fire And An Analysis Of The Programs' Representations Of Real Forests

TITLE:  Examining The Factors That Effect Random Events Through
        Coin Flips And Jingle Cups

STUDENT RESEARCHERS: Jon Bram, Abigail Neely, Nicole Warner
SCHOOL ADDRESS: Belmont High School
                221 Concord Ave.
                Belmont, MA 02178
GRADE: 11/12
TEACHER: Paul Martenis- plmartenis@aol.com 

I.  STATEMENT OF PURPOSE AND HYPOTHESIS:

We wanted to explore the theory of randomness using a coin 
flip.  We wanted to see if we could control factors that effect 
the coin flip to increase the probability of the coin landing 
heads up.  We were looking to see what factors had what effect 
on the outcome.  Our hypothesis stated that if one controlled 
enough factors they could make the outcome of a coin flip more 
predictable (the probability of getting heads up would 
increase).  

II.  METHODOLOGY:

To test our hypothesis we flipped coins, looking at different 
factors, and seeing the factors' effects on the outcome of the 
flip.

1.  We used the normal hand flip, as our control, knowing that 
the probability of landing head up was roughly even for both 
heads and tails.  Flipping it with the thumb and letting it 
land on the ground, we took the results once it was at rest.  
We repeated this 200 times.  

2. We flipped coins with our hand and eliminated the bounce.  
We eliminated the bounce by flipping the coins onto a rug to 
cushion their fall.  We did this 200 times.

3. We flipped coins using a mechanical catapult-type device.  
We did this onto a hard surface bringing the end of the 
catapult back to the same place each time.  We repeated this 
200 times.

4. We used the catapult and eliminated the bounce by letting it 
land on a soft rug.  We also only counted the coins that landed 
in a one square foot area next to the catapult.  The reason 
that we did this was that we wanted to keep the force by which 
the coin was propelled constant.  By only recording the flips 
that landed in a particular area, we kept the force constant in 
our data.  We did this time consuming process several times but 
only were able to count 98.

5. We used an electromagnet and eliminated the bounce.  We 
created the magnet with a six volt battery connected to wire 
coiled around an iron nail.  We put a washer on the magnet and 
broke the circuit so that the washer would fall the same way 
every time.  Because of time constraints we stopped after we 
got 30 "heads" in a row.

To evaluate number one and four we used a video.  We slowed 
down the video and counted the flips in the air of the trials 
with the mechanical device and several trials of the hand flip.  
We also watched bounces and counted the number of flips in each 
of those.

III.  ANALYSIS OF DATA:

Trial type           #heads     #tails     #other     %heads

1.  hand              105        95                    52.5%

2.  hand               97       103                    48.5%
(no bounce) 

3.  mechanical        102        97          1         51.0%
(start same)

4.  mechanical         88        10                    89.8%
(start same,
no bounce,
certain area)

5.  electromagnet       30        0                   100.0%
(with washer, 
from same height,
start same)

1. bounce-we found that when a flip starts random the bounce 
has no effect on the probability of heads turning up.  However, 
we found that when the flip starts controlled, but is allowed 
to bounce it becomes random. 

2. force-we thought that the mechanism (either hand, catapult, 
or electromagnet) was the most important part of the flip.  
When we combined this with the elimination of the bounce the 
probability of getting heads jumps from 50% to 90%.  And when 
we added in the reliance of the electromagnet to let the washer 
go the same way every time the probability of getting "heads" 
jumped to 100%.  The reason for this is that with the hand the 
force varies with magnitude and direction. 

3. air resistance-because we got consistent results with the 
electromagnet we felt that the effect of air resistance was 
negligible.

In our use of the slow motion video we found that the number of 
times the coin flipped in the air for the mechanical flip was 
constant, while the number of flips in the air for the hand 
flip varied.

IV.  SUMMARY AND CONCLUSION:

We found that when you control factors of a coin flip the 
result becomes more predictable.  The more factors we 
controlled the higher the probability of getting a head; since 
the probability increased our prediction of getting a head 
became more accurate.  Therefore we confirmed our hypothesis, 
the coin flip became more predictable.  We believe that by 
controlling factors that contribute to the randomness of an 
event one can make that event more predictable.

V.  APPLICATION:

This knowledge, that by controlling factors an event can become 
more predictable, can help in predicting the outcome of events.  
With the analysis of data, to see the effect certain factors 
have on an event, people will be able to better understand the 
outcomes of many events.  The application of our findings is 
much more theoretical than practical.  It is impossible to 
think that in a world such as ours one would be able to 
eliminate all of the factors that make an event random.  
Hopefully our experiment will help us to better explain and 
understand random events when we apply our method of breaking 
down all of the factors that effect the event.  



TITLE:   Is There A Connection Between The Fractal Dimension Of 
         A Rorschach Inkblot And The Usual Clinical Response?

STUDENT RESEARCHERS: Daniel Brosgol, Karen Eagar, Ealaf Majeed
SCHOOL ADDRESS: Belmont High School
                221 Concord Ave.
                Belmont, MA 02178
GRADE: 12
TEACHER: Paul Hickman- namkcihp@aol.com

I.  STATEMENT OF PURPOSE AND HYPOTHESIS:

We hypothesized that the inkblots with more common aggressive 
responses, as had been found by psychologists, would have 
higher fractal dimensions and the blots with more "placid" 
responses would have lower fractal dimensions.

II.  METHODOLOGY:

1. Find a set of the Rorschach inkblots with the common 
responses included.

2. Remove the colored inkblots (the colors in these blots 
heavily influenced the response to the blot to a point where 
making them colorless would change the initial response).

3. Scan the blots onto a computer and clean up the images.

4. Make outlines of the images or, if the outlines are the only 
images that you have available, fill in the blots.

5. Measure the fractal dimension of the blots on Fractal 
Dimension 5.1  using both the box and the fast circle method.

6. Remove inappropriate data points from the graphs.

7. Compare the fractal dimensions of the blots to see if there 
are any patterns. 
 
We want to make it clear that there are only 10 blots in all 
from the Rorschach set, 4 of which are colored.  We felt that 
the colors would have too much of an effect on the response to 
the blot.  This would lessen the effect of the FD (jaggedness) 
on the response.  We got our data concerning the responses to 
each blot from the book which had the blots in it.  The book 
used information that has been gathered by many psychologists 
over the years.

III.  ANALYSIS OF DATA:

Fractal Dimension Measurements

                            FD Full  FD Full  FD      FD
Blot     Usual Response     Blot     Blot     Coast   Coast 
                            Box      Circle   Box     Circle

1.  lions, pigs, and bears  1.83     1.75     1.41    1.57
2.  bear, gorilla           1.88     1.90     1.37    1.44
3.  animal hide or boat     1.84     1.73     1.36    1.40
4.  2 female figures        1.68     1.71     1.32    1.46
5.  bat or butterfly        1.66     1.86     1.39    1.42
6.  mask, jack-o'-lantern   1.71     1.85     1.15    1.45

The data we obtained for the coast line was not as helpful for 
our hypothesis because the measurements were so similar.  We 
based our conclusion on the data for the filled inkblots 
measured with the fast circle method, (the box method was too 
erratic to conclude anything from it). From our useful data, we 
found that images with a negative interpretation had the 
highest fractal dimensions, unlike the more passive responses 
which had the lower FD's.  

IV.  SUMMARY AND CONCLUSION:

We have accepted our hypothesis, that there is a correlation 
between the fractal dimension and the interpretation of 
Rorschach inkblots.  The correlation is that the higher the 
fractal dimension the more violent the response is.  One 
example is no. 2  which has a high fractal dimension and a 
negative response ( bear and gorilla).  The bear and gorilla 
are big animals usually feared because of their size and their 
power, making them a violent response.   An image that has a 
lower fractal dimension is no. 4.  The interpretation of it is 
two female figures facing each other.  The females have a lower 
fractal dimension, so they are responded to more passively.  
These were the two blots with the most differing FD's.

V.  APPLICATION:

We believe that someone could use this information in 
advertising because they could subconsciously elicit a certain 
response to a product that they were trying to sell.  Someone 
could use this for future psychoanalytical research by helping 
to understand why certain shapes give different responses.  
Psychoanalysts would then be able to understand the minds of 
their patients better.  This could also help those people 
taking the tests because they would know what the psychologists 
were looking for in terms of their answers.



TITLE:  The Fractal Dimensions of Sponges: Is There A
        Connection?

STUDENT RESEARCHERS:  Noah Dean Bullock, Maura Nolan Henry,
                      Joshua Charles, Asa Nissenbaum
SCHOOL ADDRESS: Belmont High School
                221 Concord Ave.
                Belmont, MA 02178
GRADE: 12
TEACHER: Paul Hickman- namkcihp@aol.com

I.  STATEMENT OF PURPOSE AND HYPOTHESIS:

We wanted to find out more about sponges and their fractal 
dimensions.  We were particularly interested in the possibility 
of a connection between the 3D fractal dimension of a sponge 
and its 2D fractal dimension.  Our hypothesis states that there 
exists a mathematical connection between the 3D fractal 
dimension of a sponge and the 2D FD of that sponge.

II.  METHODOLOGY:

Materials:

3 large masses of different sponges (cubes would be helpful)
a band saw
goggles
a balance
felt tip marker
ruler
a scanner
Fractal Dimension 5.1
Graphical Analysis

Procedure:

Be sure to dry out all of the sponges before beginning.

1. The first cube, or mass of sponge is your first iteration.  
Mass this sponge and record your data.
2. Using the felt tip marker, mark lines at half of each 
dimension of the original (height, width, length). Cut the cube 
using the bandsaw, making sure not to forget your goggles, 
along the measured lines.  The resulting cube is your second 
iteration.
3. Mass this cube and record your data.
4. Repeat steps 2 and 3 two more times, finishing with four 
individual iterations. 
5. Create a table of scale size versus mass for the four 
iterations.
6. Graph the natural log of the scale size versus the mass 
using Graphical Analysis.  This gives you your 3D fractal 
dimension.
7. Take the sponges to the scanner and scan two surfaces of 
each of them.  (This ensures accuracy.)
8. Using Fractal Dimension 5.1, measure each of the scans' 
fractal dimensions.  Use the fast circle method and place the 
center of the circle at different locations on the scan.  This 
will allow for results that better represent the whole scan.
9. The slopes of the trendlines of the graphs of your results 
reveals the 2D fractal dimensions of your scans.

One of the variables that can be controlled is the moisture 
within the sponge.  Sponges are sold wet, and so to ensure 
accuracy of mass, you must dry them out ahead of time.  Another 
variable is that of the accuracy of the sponge cut.  Some of 
the cuts may be less than accurate; this can be attributed to 
human error.

III.  ANALYSIS OF DATA:

Our first sponge had a 3D fractal dimension of 2.88 and a 2D 
fractal dimension of 1.802.  Our second sponge had a 3D fractal 
dimension of 2.98 and a 2D fractal dimension of 1.922.  Our 
third sponge had a 3D fractal dimension of 2.83 and a 2D 
fractal dimension of 1.883.

The first set of data and the third set of data are from two 3M 
sponges.  The second set of data is from one E-Z One sponge.  

IV.  SUMMARY AND CONCLUSION:

Our data do not support our hypothesis.  We did not prove that 
there is any mathematical connection between the 2D fractal 
dimension of a sponge face and the sponge's 3D fractal 
dimension.  The data we obtained shows neither a connection 
between the two brands of sponges nor one between the two 3M 
sponges.  That is not to say, though, that one does not exist.  
With more data from both other brands of sponges and more 3M 
sponges, a connection is possible.

V.  APPLICATION:

Companies which create sponges could better ascertain the most 
absorbent sponges by a combination of their material and their 
fractal dimensions.  But our research extends far beyond the 
humble walls of the modern kitchen.  If a connection between 2 
and 3 fractal dimensions is established, material engineers 
would be able to examine the porosity of ceramic building 
materials based on a side and thus be able to determine their 
strengths.



TITLE: The Effect Of Different Formation Methods On Fractal
       Dimension Of Squish Fractals

STUDENT RESEARCHERS:  Mary Tsien, Meaghan Brusch, and Steve
                      Martin
SCHOOL ADDRESS: Belmont High School
                221 Concord Ave.
                Belmont, MA 02178
GRADE: 11/12
TEACHER: Paul Martenis- plmartenis@aol.com
 
I.  STATEMENT OF PURPOSE AND HYPOTHESIS:

The purpose  of our project is to find out if the fractal 
dimension of the squish fractals using different substances is 
affected by the way in which the fractal is formed.  We will 
use four different methods: control, twist, one-corner, and 
one-side.  Our hypothesis states that the fractal dimension 
will be the same for both the control and the twist method, but 
will increase when the one-corner and one-side method are used 
due to the higher peaks which we think are formed by 
concentration on one side.  Our general hypothesis is that the 
fractal dimension will be affected by the different methods 
that are used to form the fractal.

II.  METHODOLOGY:

We are going to test our hypothesis by using four different 
methods in which to form these "squish" fractals: the control 
method (pulling straight up), the twist method (twisting to the 
right while pulling up), the one-corner method ( pulling up by 
one corner), and the one-side method (pulling up from one 
side).  We will then test these methods by using the "squish" 
method with two different substances, blue paint and Crest 
toothpaste.  We decided to use these substances because we 
thought that they would form fractals that would stay for long 
enough period of time to scan them.  The materials that we will 
need are:  several plexi-glass plates, blue paint, toothpaste, 
a scanner, a syringe, and some beakers.  We will then begin our 
experiment.  First, we will make all of the blue paint fractals 
in one day, to make sure that the paint doesn't change 
consistency.  We will then make the paint fractals by measuring 
out 1.5 mL of the blue paint in the syringe and squeezing the 
paint into the center of a  plexi-glass plate.  We then will 
place another plexi-glass plate over the substance and we will 
hold down the plexi-glass in the center for 25 seconds.  We 
then will pull the top glass off first straight up.  We will 
then repeat this control trial twice, to guarantee accuracy in 
our results, and do this for each of the four methods.  
Immediately, after we have made each fractal, we will scan it 
in the scanner to make sure that we are able to measure its 
dimension.  We will do this before the paint dries, so that it 
doesn't crack which would affect the fractal dimension.  We 
will do the same process for the toothpaste.  We will then take 
all of the scans that we have made from each substance, and 
measure the fractal dimension using the computer program, 
Fractal Dimensions 5.1.  Then we will be able to compare and 
discuss our results.

III.  ANALYSIS OF DATA:

We found that the fractal dimension increased when using the 
one-corner and one-side methods due to the higher, more 
intricate peaks that are formed.  The highest fractal dimension 
was formed using the one-side method, with an average of 1.547 
(paint) and 1.497 (paste),  which happened in both substances.  
The second highest fractal dimension was found when we used the 
one-corner method in the blue paint (avg. 1.514).  This is not 
true for our toothpaste results, but we do not think that they 
are accurate due to the fact that the peaks immediately fell 
when we formed the one-corner toothpaste fractals, making the 
dimension inaccurate.  We therefore find the blue paint 
fractals to be more accurate and will concur with the data we 
collected from that.  Unlike what we predicted, we found that 
the lowest fractal dimension was made when we used the twist 
method, average 1.300 (paint) and 1.384 (paste).  Our control 
group was the second to lowest and we used these results when 
we compared the different methods.

IV.  SUMMARY AND CONCLUSION:

We found that the fractal dimension of a squish fractal of a 
certain substance is greatest when the fractal is formed using 
the one-side method and is lowest when using the twist method.  
The dimension is highest when formed with one-side because 
there is less paint or toothpaste in one area, so there is more 
force in one area, and the paint or toothpaste gets pulled into 
more complex patterns, making the fractal dimension higher.  
The dimension is lowest when formed by the twist method 
because, when twisting, the complex patterns are smudged, 
making the dimension lower.  The one-corner method is the 
second highest in fractal dimension, a conclusion we came to 
with our paint, not toothpaste data.  This is because there is 
a greater amount of force in one area, but it is not as great 
as in the one-side method, making the dimension higher than the 
control group, but not as high as the one-side method.  

From these results, we both accept and reject our hypothesis.  
Our general hypothesis was correct, for the fractal dimension 
was affected by the method used to form the fractal.  However, 
our specific hypothesis was not totally correct.  We were 
correct when saying that the dimension would be greater using 
the one-side and one-corner method, but we were wrong when 
saying that the dimension would be the same in the control and 
twist method.  When stating this, we had not taken into account 
the "smudge" factor, and this is why the dimension of the twist 
is lower than that of the control.  Generally, we are quite 
pleased with our hypothesis. 

V.  APPLICATION:

We think that our research on fractal dimension will help the 
world by bringing  science to children of all ages who might 
not get interested in science normally.  We think that with 
this research, children's toys and posters can be made of these 
interesting pictures formed by different processes.  With this 
game, children could make their own fractals using our four 
different methods and compare them to what they resemble in 
nature.  This game would get children interested, at an earlier 
age, in learning about physics. 



TITLE: Termites: The Fractal Dimensions Of Their Random Paths

STUDENT RESEARCHERS: Brian Deese, David Millar, and Lindsay 
                     Miller
SCHOOL ADDRESS: Belmont High School
                221 Concord Ave.
                Belmont, MA 02178
GRADE: 12
TEACHER: Paul Hickman- namkcihp@aol.com

I.  STATEMENT OF PURPOSE AND HYPOTHESIS: 

We wanted to find out more about termites and the fractal 
dimension of their paths.  Does the fractal dimension of 
termite colonies increase over time as we expect?  Would the 
termite colonies resemble the aggregates and other fractal 
objects that we have studied as we expect?  Would the results 
be linear, we didn't think so.   Our hypothesis was that the 
paths that termites take, in fact, make fractal objects and 
that we can trace this fractal growth over time.  We also 
hypothesized that the growth would resemble that of the Cu 
aggregate grown in class.    

II.  METHODOLOGY:

We first acquired a movie of a growing termite colony and a 
termite computer program from B.U.  We captured pictures every 
15 frames and placed them in many different graphics programs.  
Because of the great difficulty in getting a clean, clear 
picture, we resorted to a different option.  We printed images 
of the movie out with a contrast setting of +50 in Adobe 
Photoshop.  We then outlined the path of the termites in black, 
traced the paths onto tracing paper and scanned them back into 
the computer.  We were able to measure the fractal dimension of 
these images in Fractal Dimension 5.1 and, using the fast 
circle method each time, we established specific fractal 
dimensions at each time interval.  We then plotted the fractal 
dimension with respect to time, and compared the visual result 
of the termite paths to our Cu aggregate.

III.  ANALYSIS OF DATA:

We took pictures of the termite movie at 15 frame intervals.  
We hypothesized that the fractal dimension would grow over time 
and our results are as follows:
    
# of frames             Fractal Dimension

 15                          1.259
 30                          1.314
 45                          1.458
 60                          1.487
 75                          1.548
 90                          1.597
105                          1.613
120                          1.597

From the data above, we concluded that the fractal dimension of 
paths of termites does in fact increase over time.  The graph 
leveled off at the end because our movie screen captured only a 
set portion of the termite's burrowing.  We hypothesize that 
the fractal dimension would increase more if we were to see 
more of the picture.

In regards to the termites similarity to our man-made 
aggregates, we found an interesting comparison.  The final 
fractal dimension of both were similar, but this is only due to 
the fact that we stopped the growth of both at a time when 
their fractal dimensions coincided.  There was, however, an 
interesting visual similarity in the growth of both the 
termites and the aggregate.  In the initial growth of both when 
looking at them up close, they looked surprisingly similar.  

IV.  SUMMARY AND CONCLUSION:

We found out three major things in this project: Termite paths 
are fractal objects, the fractal dimension of their paths 
increases over time as we expected in our hypothesis, and 
projects in science often become a process in how you can 
accomplish your task with the resources available.

V.  APPLICATION:

We have learned many important things in this project.  For 
research in the future, our findings can help in different 
ways.  First off, from the process that we had to go through, 
we would suggest to researchers in the future to pick a limited 
and achievable goal.  In terms of our specific research, our 
project could be a spring-board for further research.  From our 
findings, an interesting investigation would be to actually 
grow a termite colony in two or three dimensions in order to 
watch, as well as chart, the growth of the colony.  Although 
the growth of termite colonies is not the scientific gateway to 
the 21st century, knowing how these household pests' paths' 
grow could be helpful in stopping them from damaging property. 



TITLE: Heat Conduction And Its Relation To The Diffusion Of
       Particles Through A Chamber

STUDENT RESEARCHERS: Paul Adler, Colleen Baker, and Nate
                     Holloway
SCHOOL ADDRESS: Belmont High School
                221 Concord Ave.
                Belmont, MA 02178
GRADE: 11/12
TEACHER: Paul Hickman- namkcihp@aol.com

I.  STATEMENT OF PURPOSE AND HYPOTHESIS:

We wanted to compare the conduction of heat through a metal rod 
with the behavior of gas particles diffusing in a diffusion 
chamber as simulated by the program "diffusion Chamber."  We 
knew that certain aspects of the diffusing gas exhibit "random 
walk" properties and wondered if the flow of heat through a 
conductor would also have these characteristics.

Our hypothesis states that if heat was supplied to one end of a 
metal rod, the flow of heat through the rod would not take an 
amount of time linearly related to the distance it had to 
travel.  Rather, we thought that the heat would take an amount 
of time to get to the end that was exponentially related to the 
distance down the rod at which the temperature was measured.

II.  METHODOLOGY:

Equipment needed-

1.  Hot plate or other source of heat and insulating material 
(we used Styrofoam, but expect that using a better insulator 
like fiberglass would yield better results)
2.  Thermometers (we used 4)  [If electronic thermistor sensors 
were used, we suspect that more numerous and better quality 
data would be forthcoming]
3.  Metal Rod (specifically, we used aluminum)
4.  A lab stand to suspend the rod, sleeved in Styrofoam, 
horizontally with its revealed end against the hot plate, stood 
on its side.
5.  A knife to cut the Styrofoam into blocks and poke holes in 
which to insert the thermometers.

The variables in the experiment are:

1.  The dimensions and material of the rod
2.  The heat output of the hot plate
3.  Ambient room temperature
4.  Placement of the thermometers
5.  Leakage of heat past the insulation
6.  Convection of heat from the hot plate to the unshielded far 
end of the rod

We were able to control the dimensions and material of the rod, 
as well as the placement of the thermometers, but the other 
factors were not well controlled.

The procedure used in the experiment is as follows:

1. Procure an insulator such as Styrofoam and poke long holes 
through the foam to allow the metal rod to pass through as 
snugly as possible.
2. Place the rod, with the insulation surrounding it, into the 
clamp of the lab stand.  Use a ruler to mark off equal lengths 
on the insulator in order to locate the thermometers or 
sensors.
3. Place a hot plate on its side at one end of the rod, and 
adjust the clamp so that the rod-end makes a good contact with 
the face of the hot plate.
4. Make small holes in the insulation at the marked intervals, 
to allow a thermometer or sensor to protrude through the 
insulation and contact the top of the rod (it is important to 
be consistent in the side of the rod contacted, we chose the 
top since it would radiate the most heat).  Insert the 
thermometers or sensors, and ensure that a good contact is made 
(we checked the contact with our thermometers by twisting and 
observing a loud grinding sound of glass on aluminum).
5. Turn on the hot plate, and take measurements of the 
temperatures of each of the thermometers or sensors at regular 
intervals until there is a noticeable flow of heat to the end 
of the rod, where the last thermometer is placed.

III.  ANALYSIS OF DATA:

The data indicates that although the overall trend of 
temperature vs time for each distance is a linear relationship, 
the relation of distance vs time taken to reach a certain 
temperature is apparently also linear.

The data show that the temperature at the thermometer closer to 
the heat source increased much quicker than the furthest 
thermometer, however the relationship does not seem to be the 
square of the distance.  There are numerous problems with the 
consistency of this data, though; for example, the fourth 
thermometer, at 12 inches away from the heat source, seemed to 
rise faster than the third, only 9 inches away from the heat 
source.  This may be due to the fact that the end of the rod 
opposite to the hot plate is not insulated, so that the 
increase in ambient temperature caused by the hot plate and the 
convection currents it creates are heating the far end 
indirectly.  The thermometers used, too, although they were all 
made by Gita in the same factory in China, showed some minor 
discrepancy in their readings at room temperature.

In relation to our hypothesis, the data does not support our 
theory that the time required would vary in a square of 
distance relationship, as the movement of particles in gas 
diffusion chambers does.  Rather, the time required seems to 
follow a linear relationship from a cursory analysis.  The 
first, or 3-inch, thermometer took about three minutes to reach 
the 26° C mark, whereas the second thermometer, at 6 inches, 
took only five minutes to reach the same temperature.  Our 
hypothesis would predict that it should take four times as long 
to reach the same temperature at twice the distance.  This may 
be because the phenomena of metal conduction is wholly 
different from diffusion, or it may be due to the numerous 
uncontrolled factors in the experiment such as the heat 
supplied by the hot plate was not constant and there was a 
noticeable leakage of heat around the insulation by convection 
which heated the far end.  Also, we were unable to control the 
heat radiation that was emitted in large quantities from the 
surface of the hot plate.  It is more than likely that much 
heat penetrated past the first 3 or so inches of insulation and 
caused the second thermometer to rise in temperature more than 
it should if all the heat were reaching it through simple 
conduction.

IV.  SUMMARY AND CONCLUSION:

Our results in this research were not conclusive.  Based on one 
trial with poorly controlled heat supply and heat leakage, we 
found that the relationship between distance and time taken to 
reach a temperature was a linear rather than an exponential 
relationship.  This would seem to contradict our hypothesis, 
but we must remember that this was only one experiment, and it 
was far from perfect.  If we were to continue this direction of 
research, we should redesign our experimental setup, look for 
an alternate source of heat (a blowtorch, perhaps), and 
concentrate a great deal of thought on the problem of limiting 
heat leakage and convection.

V.  APPLICATION:

It is difficult to find an application for inconclusive 
results.  However, if more conclusive experiments did vindicate 
our hypothesis, we could recommend that campers worried about 
burning their hands when toasting marshmallows on metal skewers 
should purchase longer ones, since the time required to heat 
their hands will increase rapidly with longer skewers.  Cooks, 
or those using frying pans, should avoid holding the handle 
close to the pan.



TITLE:  Bumble Ball- Random Walker Or Not?

STUDENT RESEARCHERS:  Kevin P. Baratta, Deirdre A. Connolly,
                      Geetika Diddee
SCHOOL ADDRESS: Belmont High School
                221 Concord Ave.
                Belmont, MA 02178
GRADE: 11/12
TEACHER: Paul Martenis- plmartenis@aol.com 

I.  STATEMENT OF PURPOSE AND HYPOTHESIS:

We wanted to find out more about the path of the bumble ball.  
We hypothesized that the bumble ball behaves as a random 
walker.

II.  METHODOLOGY:

We tested our hypothesis using the bumble ball, chalk, a stop 
watch, a paper grid to collect data on, blue and red pens to 
mark the path of the bumble ball, and paper labels A-I.  We 
composed a sixteen by sixteen square grid on the floor.  We 
traced the path of the bumble ball on the paper grid and 
dropped labels marking the position every ten seconds.   We ran 
this process for ninety seconds per trial and performed as many 
trials as we had time for.  We ran the Many Walkers program for 
the purpose of comparing the average X 2.

III.  ANALYSIS OF DATA:

Our data indicated that the bumble ball was a random walker.  
We recorded the location of the bumble ball every ten seconds 
for ninety seconds.  We analyzed this data in two ways.  First, 
we plotted the locations of the bumble ball for each trial at 
every ten second interval on nine separate grids.  We found 
that the points were not distributed to any particular side and 
there were very few if any repeating points.  This indicated 
that if the process was run more than once under identical 
circumstances, the results would not be the same and thus 
unpredictable.  This means that the bumble ball is a random 
walker.

We measured the distance from the starting point of the bumble 
ball's path to each of its locations at the ten second 
intervals.  After collecting the distances for each location A-
I, over forty trials, we composed an Excel  spread sheet to 
find the overall average squared value at each distance.  In 
finding this value, we were able to compare our results to the 
results of the many walker simulation.  The graphs of the 
bumble ball and many walkers are similar for the same number of 
walkers and data points.    Both sets of data points fall close 
to a line, but not on it, due to the limited number of trials 
we were able to perform.

IV.  SUMMARY AND CONCLUSION:

Through the analysis of our data, we found that the bumble ball 
is a random walker, which is in accordance with our hypothesis.  
After forty trials of performing the experiment, we were unable 
to predict the location of the bumble ball at any particular 
time.  

V.  APPLICATION:

Our project gave us a "hands-on experience" in the world of 
Randomness. The method we used can be applied to study other 
mechanisms (toys) like the bumble ball. 



TITLE:  Comparison Of The Growth Patterns In Forest And Fire
        And An Analysis Of The Programs' Representations Of
        Real Forests

STUDENT RESEARCHERS: Boris Klimovitsky, Erik Landfried, Kaitlin
                     McGaw
SCHOOL ADDRESS: Belmont High School
                221 Concord Ave.
                Belmont, MA 02178
GRADE: 11/12
TEACHER: Paul Hickman- namkcihp@aol.com

I.  STATEMENT OF PURPOSE AND HYPOTHESIS:

We want to explore the different ways of growing a forest, 
using the programs Forest and Fire. We want to determine which 
program more accurately portrays a true forest, as well as the 
differences between the two methods.  We think that Forest will 
better represent an authentic forest, as it grows from the 
center out with each seed's placement dependent on the growth 
of the previous; whereas, Fire  grows in rows, regardless of 
previous seeding.  We also believe that the two methods should 
produce dissimilar fractal dimensions. 

II.  METHODOLOGY:

The first thing we did was explore the two programs 
extensively.  We attempted to look for the possible quirks of 
each density.  To compare the programs, we found the fractal 
dimensions of each at the same density and compared the values.  
The method to find each fractal dimension is as follows:

1) Create a forest picture by clicking on the appropriate 
density.
2) Capture this image.  (For Fire , simply select "Save 
Picture." from File menu.
3) Open MacPaint,  and paste the captured image into a file; 
select "Invert" to switch coloring for proper scan.
4) Open Fractal Dimension 5.1 , and then open the MP file you 
just created.
5) Measure the image using both the box and circle methods.
6) Look at the graph, and clean it up appropriately, 
eliminating points which are not on the line.
7) The value for the slope of this line is the fractal 
dimension.
8) Compare the fractal dimensions for both programs to see if 
the method of growing the forest affects the fractal dimension.

III.  ANALYSIS OF DATA:

The formation of forests is based on an intricate chain of 
events, including wind, topography, climate, earth history, and 
soil development.  Real forests also start out from a central 
seed and grow outward until they reach a stage called the 
climax forest, which is the point at which little change 
occurs.  Climax forests have no new types of trees growing 
there, and there is minimal change in the overall shape of the 
forest.  Climax forests perpetuate endlessly.  ( Farb, Peter. 
The Forest. New York: Time-Life Books, 1963)

IV.  SUMMARY AND CONCLUSION:

In order to make a comparison between the two programs, we 
sought to find the fractal dimension of the forests that each 
created.  We decided to look at the forests created at the 
critical probability, when there is a 50% chance that the 
forest will percolate.  Percolation is the growth of a fractal 
design from one side to the other.  In our case, we examined 
images where the forest reached two sides of the grid to which 
it was limited.

We decided to examine Forest at .5 and .6 since we were unable 
to look at its critical probability of .59.  We looked at one 
forest grown at each of the two probabilities.  We found the 
fractal dimensions to be for .5 and .6, 1.321 and 1.431, 
respectively.  For Fire, we found that the fractal dimension 
was 1.668.  For the three sample forests we measured, the shape 
made by the Fire program had a higher fractal dimension.  
Although we measured only three sample forests, we proved our 
hypothesis that the different methods would produce dissimilar 
fractal dimensions.

We also found that both programs have similarities with real 
forests, yet the differences clearly outweigh these few 
similarities.  Forest, we believe, represents an individual 
forest.  It starts in the center and grows out from there until 
something curbs its growth (amount of moisture, poor soil, 
rock). It also realistically reaches a state of "climax 
forest", where there are minimal future changes (on the program 
the final forest is surrounded by rocks).  The forest in Fire, 
however, seems like it reaches beyond the grid endlessly.  The 
method of growth for Fire is inaccurate in that forests simply 
do not grow from a single seed, independent from surroundings 
and previous growth.  Neither program accounts for the forces 
of nature that affect a real forest, such as wind, topography, 
climate, soil development, etc.  We have decided that Forest 
more accurately represents a true forest.

V.  APPLICATION:

We think that the two programs are excellent for the unit on 
fractals, but fail to accurately portray true forests for 
students to examine.  A better program, including a combination 
of Forest as well as the natural factors which affect a forest, 
would be a more useful tool for learning about forests and 
natural fractals.

© 1997 John I. Swang, Ph.D.