Projects:2018s1-141 CSI Adelaide: Who killed the Somerton Man?
Contents
Supervisors
Honours students
Abstract
The corpse of a mystery man’s body was found at Adelaide’s Somerton Beach, South Australia, Australia, on the 1st December 1948 and was hence referred to as the Somerton Man. Till this day the identity of the Somerton Man and the cause of his death is unknown. This project will be broken up into three different tasks, with all contributing towards the unsolved case of the mysterious man. The first task is based around the piece of paper that was found inside his trousers pocket. This piece of paper had five lines of capital letters, and is thought to be some kind of code or cipher. It was found that this mysterious code was a part of a poetry book, known as the ‘Rubaiyat of Omar Khayyam’. The letters on the mysterious code are thought to be first letters of words, based on previous year’s project groups. The location of his death is near Morphettville Racecourse, which leads to belief that the Somerton Man’s mysterious code are different horse names. However, using various statistical approaches, this was proven to be not true. The second part of this project involved, using a mass spectrometer to analyse different isotopic signatures of the samples. More specifically, the samples used were the shaft of the hair, obtained from various people, which were ablated by the laser, and effectively recorded the various elements. This will be compared with the Somerton Man’s hair, to identify specific elements present, as well as noting how long he was in Adelaide before his death. Different DNA samples, were analysed in the final task. Using software tools the samples were degraded, until the DNA became unidentifiable. It was found that degrading 50% or greater of the DNA sequences, makes the DNA unidentifiable.
Acknowledgements
Project supervisor Professor Derek Abbott, for the helpful and motivational advice, as well as the exceptional guidance, which was presented on each of the completed tasks. The University of Adelaide Microscopy Centre, for permitting the project team to use the mass spectrometer. Dr Sarah Gilbert for the assistance and suggestions whilst using the mass spectrometer. Also, thank you to all the volunteer who gave their hairs to use during the experiment. In addition, the project team would like to acknowledge all the software support from University of Adelaide, including Microsoft, Matlab as well as Iolite from the Microscopy centre.
Introduction
Motivation
During this project, a study was undertaken on an unsolved murder case. On December 1st, 1948, an unknown man was found deceased, on Somerton Beach in Adelaide [1]. From then on, he was labelled, ‘the Somerton Man’. There was no form of identification present on him, as well as little information surrounding his death. Therefore, the cause of his death is still unknown to this day [2]. Figure 1 shows the deceased man.
A piece of paper with the words “Tamám Shud” printed on it, was found rolled up inside his trousers pocket, which can be seen in Figure 2. This statement can be translated from Persian to either “it is ended” or “it is finished”. This piece of paper was found to be part of a poetry book, the ‘Rubaiyat of Omar Khayyam’ [3]. The theme of the book is that, one should live life to the full and have no regrets when it ends [4].
The book is speculated to be related to the dead man, due to the parchment. Hence, the case being known as the Tamam Shud case. This has been considered, since the early stages of the police investigation, to be "one of Australia's most profound mysteries [6].” Capital letters were found to be scribbled in the back cover of the Rubaiyat of Omar Khayyam, as seen in Figure 3. Thus, indicating that these letters are somewhat vital to the case at hand, as it is speculated that they may be a form of code or cipher.
The code consisted of five various lines of capital letters, with a total of fifty letters all together. The second and fourth lines are very similar in the way in which they appear. It is therefore, believed that the Somerton Man may have made an error with the encryption, when writing the second line, hence why it is struck out. It is unclear whether some of the letters are in fact an “M” or “W” [7]. For tasks referring to the code, it is assumed that the unclear letter is an “M”. His body was found to be located near the Morphettville Racecourse, this leads to the belief, that the Somerton Man’s mysterious code are related to different horse names. It was also noted that the people who found the deceased body of the Somerton Man, were racehorse jockeys [2]. An Inductively Coupled Plasma Mass Spectrometer (ICP-MS) is used in this project to identify the various isotopes in various people’s hair. With the main concern of Strontium as, Adelaide has high levels in strontium in the soil compared to the rest of Australia.
Previous Studies/Related Work
Australian Department of Defence
In 1978, a request was sent by journalist Stuart Littlemore to the Department of Defence cryptographers to analyse the code. Unfortunately, the cryptographers were unable to crack the code, as they could not provide a satisfactory answer. It was stated that the code had “insufficient symbols” or a “disturbed mind” generated the meaningless code [8].
The University of Adelaide project groups
There have been several Honours project groups at the University of Adelaide that have undertaken this project. The previous work the project groups have done include:
- Letter frequency analysis in different languages.
- Initial letter and sentence letter probabilities.
- The likelihood of the code being an initialism of a poem.
- Different cypher techniques.
- The design and implementation of a web crawler.
- 3D generated reconstruction bust of the Somerton Man.
Main conclusions that these project groups have come to are:
- It is unlikely that the mysterious code is created randomly.
- There is strong evidence to believe the mysterious code is most likely to be in English.
- It is not likely that the mysterious code are initialisms extracted from poems.
- The Rubaiyat of Omar Khayyam was not used as a straight substitution one-time pad for encryption.
- The Rubaiyat of Omar Khayyam was not created as a one-time pad for the mysterious code.
With these conclusions, this project will look into further detail of what the Mysterious Code is [9] [10] [11] [12] [13] [14] [15].
Mass Spectrometer Pervious Work
Previous years have also done study with Mass Spectrometer. The 2013 project group had some of Somerton Mans hair and plotted the different elements in the hair comparing with controlled samples. Analysis was conducted on different elements between the two samples. This was done using a glass slide, which have impurities in it [13]. In the 2016 project group, they recreated the analysis using a quartz slide, which does not have impurities. They concluded that Somerton Mans had some abnormally high readings, of some elements, one of which is strontium [15]. In this project, the strontium level will be looked in higher detail and this will be used to indicate how long the Somerton Man was in Adelaide before his death.
Aims and Objectives
The first task that will be done is to understand if the mysterious code are a collective object (horse names, Adelaide street names, Australian beaches, etc.). This will be done using hypothesis testing. An extension of this task will also be completed, which involves the Rubaiyat of Omar Khayyam. The second task that will be performed is using a mass spectrometer. Controlled sample hairs will be compared with Somerton man’s hair, to see how long he was in Adelaide before his death, by finding different elements in the hair. The third task that will be accomplished is using DNA data. The data will be degraded using software tools till it becomes unidentifiable. This then can be used to see how much DNA we need from Somerton man, where further research can undergo.
Technical Background
P-value Theorem
A p-value is a recognised statistical probability, which acknowledges whether an equal or larger effect is present, in comparison to its observed counterpart. In statistics, the p-value helps you determine the significance of the statistical hypothesis by observing the results that were sampled. This determines the probability that the results are due to chance, rather than the experimental conditions. Thus, determining the strength and validity of the results against that of the null hypothesis [16]. In this project, the main focus with the p-values is to determine whether the mysterious code, represents local horse names. Where the null hypothesis is ‘The group of letters are horse names’ with the alternative hypothesis being ‘The group of letters are not horse names.’ For null hypothesis to be accepted the p-value must be larger than 0.05, this indicates that the observed data point is located in the ‘most likely observation’ range, as seen in Figure 4. If the p-value is lower than 0.05, this indicates that the collected results are statistically significant and that the observed data point is located in the ‘very unlikely observations’ range. If the results are in the ‘very unlikely observations’ range, then the null hypothesis can be rejected, which means that the mysterious code is indeed, not horse names [17].
Figure 4 shows, the y-axis is the probability and the x-axis is the set of possible results.
Mass Spectrometer
A mass spectrometer is an analytical technique which, when given a foreign sample, it can detect unknown compounds within it. The mass spectrometer produces a multitude of ions in the sample, which then uses a mass to charge ratio of the different ions and records the quantity of each ion type [19]. The components of a typical mass spectrometer is seen in Figure 5. The three major components are the ion source, analyser and the ion detector system.
The ion source produces gaseous ions from the sample that was used. The analyser, then sorts the different ions by using the mass-to-charge ratio, according to their mass components. The ion detector system detects the different ions in the sample and records the quantity of each ion type and converts it into an electric signal [20]. The Inductively Coupled Plasma Mass Spectrometer (ICP-MS) is the type of mass spectrometer that was used in this project. The ICP-MS is faster, more precise and sensitive at finding different ions compared to other types of mass spectrometer [21]. In regards to the project, the sample being used is the shaft of the hair.
Single Nucleotide Polymorphism (SNP)
Single Nucleotide Polymorphisms (SNPs), are one of the most common and well talked about genetic variations, which can be present between humans [22]. These variations occur within a nucleotide (a single block of DNA) and happen roughly within 1 out of 300 base pairings [23]. In regards to the project, SNPs will be removed from a DNA sample.
Knowledge Gaps and Technical Challenges
The technical challenges that will be encountered in this project are related to all the knowledge gaps mentioned. To complete each task within the project, further development for programming skills, such as Matlab were required. P-value calculation and hypothesis testing needed some revision, to ensure that a satisfactory level of understanding of the concepts was present. The skill to use Microsoft Excel to perform statistical analysis on the p-value, is required. It was also required to learn how to correctly use a mass spectrometer and interpret the results, this can be done by finding multiple ways to enhance knowledge, before trials.
Task 1: Code Analysis
Aim
The aim of this task is to comprehend whether the mysterious code represents some collective object. The collective objects that will be used are horse names, Australian beaches and cities, South Australian street names, and The Rubaiyat of Omar Khayyam book. The assumption will be made that the letters in the mysterious code, are the initialism of a word. The Somerton Man had a lot of associations with horses, so further research will be going into the assumption that the mysterious code are ‘horse names’. The null hypothesis is ‘The group of letters are horse names’ and the alternative hypothesis is ‘The group of letters are not horse name.’
Preperation
Before the initialisation of this task, the details of the whole case were reviewed. This review included the following; who found the corpse, where it was located and when it was discovered. This gave a better understanding of what type of collective object should be researched, which are stated above. As mentioned before, the reason why horse names are a major factor is because the location of his body was found to be located near Morphettville Race course, as seen in Figure 6.
It should also be noted that the people who found the deceased body were racehorse jockeys. Hence the reason we assumed the mysterious code are the initialism of horse names from the year 1948.
Method
The approach to determine if the mysterious code represents collective objects, will be done by calculating the p-value and implementing hypothesis testing. In the case of the horse names, there were no direct websites, which provided horse names in the year of 1948. This led to discovering evidence of these names within relevant newspapers and articles. This was done by using ‘Trove’, a search engine to help find resources in Australia. More specifically, in this case it was used to obtain articles and newspapers from 1948. The other collective objects, as mentioned above, were found using South Australian government websites. This led to an abundance of cross checking, to make sure that the list which was going to be utilised, was indeed correct. Matlab was used as the software tool. The initial letter of the each collective object was all that was needed, where using code was implemented to perform this task. In the case, where a collective object had multiple words, both words were included. Also if there was any extra punctuation, it was removed. A demonstration is shown in Figure 7, which illustrates the input and output results for the horse names case and also displays the frequency of each letter.
Excel was used, to produce the statistical results from Matlab. The letter frequency will be obtained by dividing the amount of each letter over the total amount calculated, where then a p-value test was performed and a comparison graph was completed.
Results
Each collective object was compared to the mysterious code by the frequency of each letter. Where the x-axis represents the alphabet and the y-axis represents the frequency of the letters between the two testing objects. The p-value test was also completed to verify the results, where a p-value of less than 0.05 shows that it is very unlikely that the collective object is the mysterious code.
Horse Names
The comparison of horse names to the mysterious code is seen in Figure 8.
There was a sample of 69 horse names and it can be seen on the graph that the horse names do not correlate with the mysterious code with many of the English letters. This was also proven by the p-value, as it was lower than 0.05, which means the null hypothesis is not accepted.
Australian Beaches
The comparison of Australian beaches to the mysterious code is seen in Figure 9.
There was a sample of 114 beach names. Analysing the graph it be seen that the frequency of the letters do correlate with mysterious code. As the results seemed genuine a hypothesis test was done between this values. The results showed a p-value of greater than 0.05, which indicates that the mysterious code could be Australian beach names.
South Australia Street Names
The comparison of South Australian street names to the mysterious code is seen in Figure 10.
There was a sample of 447 South Australian street names. Observing the graph it can be seen that the frequency of the letter are not similar with the mysterious code. This was also proven by the p-value, as it was lower than 0.05.
Australian City's
The comparison of Australian city names to the mysterious code is seen in Figure 11.
There was a sample of 90 Australian city names. Observing the graph it can be seen that the frequency of some letter are similar with the mysterious code. A hypothesis test was then done to check the results. The p-value that was obtained was less than 0.05.
The Rubaiyat of Omar Khayyam book
The comparison of the Rubaiyat of Omar Khayyam book to the mysterious code is seen in Figure 12.
There was a sample of 852 words form the book. Observing the graph it can be seen that the frequency of the letter are not similar with the mysterious code. This was also proven by the p-value, as it was lower than 0.05. An extension of this task was also done. This includes analysing The Rubaiyat of Omar Khayyam book more carefully. Previous years stated that the mysterious code does not correlate with the book. Each paragraph in the book has four lines of words (see Figure 13), which compared with the mysterious code also has four lines. Still assuming that each letter in the mysterious code is an initial word, we can compare the two.
The task was to count how many words are in each line of the book and compare it with the mysterious code. Using the first paragraph in Figure 13 (outlined with a red square), the first line has 9 words, then followed by 7 words in the second line, then 8 words in the last two lines. Comparing just the first paragraph with the mysterious code from line 1 to 4, there are 9, 11, 11 and 13 letters respectively. It already can be seen from the first paragraph that there may not be a correlation between the mysterious code and the book. Counting every line would be very time consuming, therefore a text file of The Rubaiyat of Omar Khayyam was used, in correlation with Matlab to count each word in each line. Then using excel, a graph was plotted with error bars to the number of letters in the mysterious code. This can be seen in Figure 14.
The x-axis represents which line in the paragraph it is and the y-axis represents the amount of words present in that line. It can be seen that on line 1, the mysterious code is in the error bars. The rest of the lines are out of the error bars. This indicates that the mysterious code is not from The Rubaiyat of Omar Khayyam book and further proves the previous year’s studies of the book not being part of the mysterious code.
P-value
A summary of the p-values are shown in Table 1.
It can be observed that the only collective object that is above 0.05 is Australian beaches, this indicates that the mysterious code could be Australian beaches.
Conclusion
Overall, the results have shown, that it is unlikely that the mysterious code represents the hypothesis stated, of that it is horse names. It has also shown that it is unlikely to be South Australian street names, Australian city names or The Rubaiyat of Omar Khayyam book, even with extensive analysis on the book. But to some surprise it is possible that the mysterious code is Australian beach names, as the p-value was above 0.05. Future research will be required to obtain more meaningful results.
Task 2 Hair Analysis
Background Theory
Analysing elements in the hair could reveal a great deal of useful information on a person’s recent life, this includes their lifestyle imbalances, living environment and dietary problems. Also the mineral levels in hair is about ten times more robust, than compared to blood [26]. These results could reveal where the last place Somerton Man had been to or even the last activity that Somerton Man had done, which could provide some useful evidences in solving this case. Knowing how much hair grows is very important for this project. For every month that passes, hair grows by 1cm [27]. This means the newest hair is in the root.
Aim
The aim of this task is to identify the different isotopes present in several different people’s hair. More specifically the element of concern is strontium. Adelaide has high levels in strontium in the soil compared to the rest of Australia. With this knowledge, the task is to test various hair samples, which have left Adelaide, within the past month and compare it to that of hair samples that have not left Adelaide for at least a year, to see how the strontium values change. This will then be compared with the strontium levels in the Somerton Man’s hair, which can determine how long he was in Adelaide before his death. The ICP-MS, is the approach to determine the different isotopes within the hair, which will then return a spectral analysis of the hair. The spectral analysis will be completed by laser ablation of the hair, where the hair will be ablated with a laser and the spectral elements are recorded.
Preparation
Hair Elements
Before the ICP-MS could be used for the analysis on the hair samples, different isotopes had to be chosen for the ICP-MS to find. Research was completed to find the most common chemical elements inside human hair, which included carbon, hydrogen, oxygen, sulphur, phosphorus and zinc. Then further research to indicate what element we want the ICP-MS to find, which included any toxic elements and common elements in food. There were 24 isotopes that were recorded by the ICP-MS and are shown in Table 2.
Hair Samples
In this project, a different hair samples was obtained, from five different individuals, for more reliable and accurate results. Two of these samples obtained, were from people that left Adelaide for a period of time and the other three samples were from people that stayed in Adelaide. All the samples will be anonymous for privacy reasons and will be identified as A, B, C, D and E. A summary of each hair sample is shown in Table 3, this includes which sample is identified as, sex, the date the hair was obtained and a brief description of the sample.
The length of time out of Adelaide is very important, as we can determine the length of hair that was in Adelaide and the length of the hair that was outside of Adelaide. Sample A left Adelaide for 7 days, assuming there is 30 days in one month and hair grows 1cm per month, therefore 2.33mm of hair is from Japan. Similarly with sample E, they left Adelaide for 13 days, therefore 4.33mm of hair is from Bali. Also have to take consideration of the 2 days from when they came back to Adelaide to when the hair was obtained, this is 0.67mm. An example of the length of the hairs is shown in Figure 15.
The red displays the 2 days before obtaining the hair, when both samples came back to Adelaide. The orange shows the length of sample A and the green shows the length of sample E, when both samples were away from Adelaide.
