Texarkana Moonlight Murders: The Phantom on the Border of Two States

Symbols, Traces, My Case Investigation Record 39

Texas Canada Moonlight Murder
Phantom of the Two States

In the spring of 1946, a lover and a couple were attacked one after another in Texas and Arkansas, a city on the border between Texas and Arkansas. The media called the unidentified criminal a "phantom killer."

Figure 1. Texas Canada separating Texas from Arkansas — Image:

From the conclusion of this record to
It is likely that one central criminal was involved in several attacks, and Eur Sweeney is a suspect with great explanatory power. However, beyond the spouse's statements and circumstances, the case cannot be closed under his name because there is no material evidence verified in the trial.

Prologue: Recorded Night

After the night deepens, the data was spread on the desk. The first thing that caught my eye was not the answer, but the imbalance. Some scenes were as clear as photographs and some time zones were empty as a whole. People try to capture one clear scene and explain the whole thing, but I marked the empty section first. Mysteries arise because there are many clues, but they also arise because the distance between clues is too far.

The first three attacks took place on remote roads or places sought by lovers, and the last one was directed at farmhouses. In several cases, .32 caliber pistols of the same series were treated as key links. The FBI has released vast investigative files but no one has been convicted of murder.

I deliberately read the famous hypothesis at the end. First of all, when we encounter the conclusion, it seems that all objects and symbols exist for that conclusion. Therefore, the timetable, place, object, and original text were organized first, and the witness and interpretation were placed in different compartments. The novel-like atmosphere was drawn from the cold texture of the actual record, not from a made-up secret.

Figure 2. Lover Road outside 1946 — Image:

Case File: What I Could Believe

20313233343536373839404142434445464748495051525354555657585960616263646566212223242526272829303132333435363738394041424344454647484950515253545556

Number	Confirmed Record	Current Reliability
1	From February to May 1946, four attacks continued in the Texarkana area.	High
2	Five people died and three survived the attack.	High
3	The same ballistic characteristics of several events.It pointed to a .32 caliber automatic pistol series.	High
4	The site of the attack spanned roads to and from the state border and outlying farms.	Medium
5	Url Sweeney is on the main suspect list, but he has not been tried on murder charges.	Medium

Whether all of the attacks were the same offender, the whereabouts of the used pistol, and the strength of non-disclosure evidence against Eur Sweeney have not been confirmed.

Caution trap
'Unexplained' is not the same as 'supernatural', 'a huge conspiracy', and 'only this hypothesis is possible'. I didn't count the blanks as evidence.

Figure 3. Reproduction of the white cloth mask mentioned by the witness — Image:

Scene Reconstruction: Finally, the place of the certain point

was changed to the actual movement, not the stage in your head. We divided the starting point, the last check point, and the location where the item was found and calculated the minimum time required in between. When the dramatic description of later generations was removed, small and specific elements such as sound and lighting, doors and roads, paper and ink, and the spacing of symbols came forward.

The person's mind was not fixed and the action was written first. The same behavior can be caused by fear, planning, error, and fatigue. If it is a code, the same symbol can be a sound, a word, a separator, and a meaningless sign. After writing down all possible roles, he erased the roles that conflict with other records one by one.

Clue 1. Four attacks continued in the Texas area from February to May 1946.

I split this sentence into three pieces first. It indicated who first checked, how close the confirmation was to the incident, and where the later interpretation was attached. Old events and codes turn into clearer stories than the original while being repeatedly cited. But clarity is not the same thing as accuracy. In my notes, I separated the direct record, the traces left by the object, and the reasoning of later generations into different colors.

Then, the clue was matched with the record, "Five people died and three survived in the attack." It is not a good hypothesis if one hypothesis explains only the first sentence and avoids the second sentence. In particular, the attractive explanation also lowered the score if it failed to pass the question, "Is the change in the method of the last attack the same criminal's adaptation or the trace of a different criminal?" I've seen clues that make the conclusion uncomfortable longer than clues that make the conclusion cool.

If you look at the clues this way, the case seems a little less dramatic. Instead, there is a boundary between how far it is clear and where it is imagined. That boundary is necessary to correct reasoning when new data are released. The most dangerous moment in the unwritten record is not when you make a false hypothesis, but when you turn the hypothesis into a sentence of fact.

Clue 2. Five people died and three survived the attack.

I split this sentence into three pieces first. It indicated who first checked, how close the confirmation was to the incident, and where the later interpretation was attached. Old events and codes turn into clearer stories than the original while being repeatedly cited. But clarity is not the same thing as accuracy. In my notes, I separated the direct record, the traces left by the object, and the reasoning of later generations into different colors.

Then, this clue is "the same ballistic characteristics of several events."It pointed to the .32 caliber automatic pistol series." It is not a good hypothesis if one hypothesis explains only the first sentence and avoids the second sentence. In particular, if you fail to pass the question, "How exactly does the cycle of stolen vehicles overlap with the date of attack?" the attractive explanation also lowered the score. I've seen clues that make the conclusion uncomfortable longer than clues that make the conclusion cool.

If you look at the clues this way, the case seems a little less dramatic. Instead, there is a boundary between how far it is clear and where it is imagined. That boundary is necessary to correct reasoning when new data are released. The most dangerous moment in the unwritten record is not when you make a false hypothesis, but when you turn the hypothesis into a sentence of fact.

Figure 4.32 Caliber automatic pistol and bullet — Image:

Clue 3. The same ballistic characteristics of several events.It pointed to a .32 caliber automatic pistol series.

I split this sentence into three pieces first. It indicated who first checked, how close the confirmation was to the incident, and where the later interpretation was attached. Old events and codes turn into clearer stories than the original while being repeatedly cited. But clarity is not the same thing as accuracy. In my notes, I separated the direct record, the traces left by the object, and the reasoning of later generations into different colors.

This clue was then confronted with the record, "The site of the attack was across roads and outlying farms to and from the state border." It is not a good hypothesis if one hypothesis explains only the first sentence and avoids the second sentence. In particular, if you fail to pass the question, "Does the description of the survivor's mask refer to the same person?" the attractive explanation also lowered the score. I've seen clues that make the conclusion uncomfortable longer than clues that make the conclusion cool.

If you look at the clues this way, the case seems a little less dramatic. Instead, there is a boundary between how far it is clear and where it is imagined. That boundary is necessary to correct reasoning when new data are released. The most dangerous moment in the unwritten record is not when you make a false hypothesis, but when you turn the hypothesis into a sentence of fact.

Clue 4. The site of the attack spanned roads to and from the state border and outlying farmhouses.

I split this sentence into three pieces first. It indicated who first checked, how close the confirmation was to the incident, and where the later interpretation was attached. Old events and codes turn into clearer stories than the original while being repeatedly cited. But clarity is not the same thing as accuracy. In my notes, I separated the direct record, the traces left by the object, and the reasoning of later generations into different colors.

This clue was then matched with the record, "Url Sweeney was on the main suspect line, but he was not tried for murder." It is not a good hypothesis if one hypothesis explains only the first sentence and avoids the second sentence. In particular, the attractive explanation also lowered the score if it failed to pass the question, "Is the change in the method of the last attack the same criminal's adaptation or the trace of a different criminal?" I've seen clues that make the conclusion uncomfortable longer than clues that make the conclusion cool.

If you look at the clues this way, the case seems a little less dramatic. Instead, there is a boundary between how far it is clear and where it is imagined. That boundary is necessary to correct reasoning when new data are released. The most dangerous moment in the unwritten record is not when you make a false hypothesis, but when you turn the hypothesis into a sentence of fact.

Clue 5. Your Sweeney is on the main suspect list, but he has not been tried for murder.

I split this sentence into three pieces first. It indicated who first checked, how close the confirmation was to the incident, and where the later interpretation was attached. Old events and codes turn into clearer stories than the original while being repeatedly cited. But clarity is not the same thing as accuracy. In my notes, I separated the direct record, the traces left by the object, and the reasoning of later generations into different colors.

This clue was then matched with the record, "Four attacks continued in the Texas area from February to May 1946." It is not a good hypothesis if one hypothesis explains only the first sentence and avoids the second sentence. In particular, if you fail to pass the question, "How exactly does the cycle of stolen vehicles overlap with the date of attack?" the attractive explanation also lowered the score. I've seen clues that make the conclusion uncomfortable longer than clues that make the conclusion cool.

If you look at the clues this way, the case seems a little less dramatic. Instead, there is a boundary between how far it is clear and where it is imagined. That boundary is necessary to correct reasoning when new data are released. The most dangerous moment in the unwritten record is not when you make a false hypothesis, but when you turn the hypothesis into a sentence of fact.

Figure 5. The flow of stolen and abandoned vehicles — Image:

Put three hypotheses on the same scale The

hypothesis was not evaluated as a cool sentence. The number of clues to be explained, the number of newly requested assumptions, and the degree to which they endure the opposite evidence were written, respectively. Even if one clue is fully explained, it is not a good hypothesis if the other records have to be discarded.

Hypothesis A. The same criminal launched a series of attacks, changing places and targets.

This explanation stated, "Four attacks continued in the Texarkana area from February to May 1946."It has the advantage of accepting " as a relatively small assumption. I assumed that this hypothesis was correct and walked time and space all over again. Each time people, objects, and symbols moved to the next location, they wrote down the actions and times they needed. There must have been a hidden assumption at the point where the sentence was cut off.

The biggest weakness is "the same ballistic characteristics of multiple events."It refers to the .32 caliber automatic pistol series." If you keep adding accomplices, coincidences, and lost things to keep the hypothesis alive, you can explain anything. However, such an explanation does not predict new clues. I only scored when two independent records pointed in the same direction, and I scored every time I called in a new exception.

117100118

So, whether or not to adopt this hypothesis as a conclusion was judged not by atmosphere but by cost. The number of clues to be explained, the number of newly requested households, and the degree to which they endure evidence against them are listed in the same table. Novel scenes were used to show the texture of facts, but not as evidence to fill in the blanks.

119101120

Hypothesis B. The first three cases are one criminal and the last farmhouse attack is an imitation or separate crime.

121102122

This explanation has the advantage of accepting "five people died and three survived the attack" as a relatively small assumption. I assumed that this hypothesis was correct and walked time and space all over again. Each time people, objects, and symbols moved to the next location, they wrote down the actions and times they needed. There must have been a hidden assumption at the point where the sentence was cut off.

123103124

The biggest weakness is that it does not fully explain "the site of the attack was across roads and outlying farms to and from the state border." If you keep adding accomplices, coincidences, and lost things to keep the hypothesis alive, you can explain anything. However, such an explanation does not predict new clues. I only scored when two independent records pointed in the same direction, and I scored every time I called in a new exception.

125104126

So, whether or not to adopt this hypothesis as a conclusion was judged not by atmosphere but by cost. The number of clues to be explained, the number of newly requested households, and the degree to which they endure evidence against them are listed in the same table. Novel scenes were used to show the texture of facts, but not as evidence to fill in the blanks.

127105128

Hypothesis C. A local criminal using a stolen vehicle serialized an accidental attack.

129106130

This explanation is "the same ballistic characteristics of multiple events."It pointed to a .32 caliber automatic pistol series.It has the advantage of accepting " as a relatively small assumption. I assumed that this hypothesis was correct and walked time and space all over again. Each time people, objects, and symbols moved to the next location, they wrote down the actions and times they needed. There must have been a hidden assumption at the point where the sentence was cut off.

131107132

The biggest weakness is that it does not fully explain "Url Sweeney is on the main suspect line, but he has not been tried for murder." If you keep adding accomplices, coincidences, and lost things to keep the hypothesis alive, you can explain anything. However, such an explanation does not predict new clues. I only scored when two independent records pointed in the same direction, and I scored every time I called in a new exception.

133108134

So, whether or not to adopt this hypothesis as a conclusion was judged not by atmosphere but by cost. The number of clues to be explained, the number of newly requested households, and the degree to which they endure evidence against them are listed in the same table. Novel scenes were used to show the texture of facts, but not as evidence to fill in the blanks.

Figure 6. The outer farmhouse where the last attack took place — Image:

Conclusions I abandoned

137111138

When I started the investigation, I was also drawn to the most famous explanation. It organized the case into one sentence and gave intention to everything. However, in contrast to the original record, each blank required a new coincidence. I erased the conclusion then and started again. Good reasoning is not about adding anything, but about removing unnecessary assumptions.

139112140

The ballistic connection is much stronger than the 'phantom' made by the media. But the same pistol doesn't mean the same hand. I scored the location, vehicle, and ammunition separately, and the continuity of the first three attacks was considered high, but the last farm case needed to be verified separately.

141113142

In particular, the details repeated on later articles and the Internet were separated from the source. Just because the same sentence appears in multiple places does not mean that there are multiple independent sources. We first checked the possibility that one error was duplicated. After this process, the case became less colorful, but the weight of the remaining clues increased.

Figure 7. A joint investigation between the Texas Rangers and the local police — Image:

Anti-evidence: Sentence to destroy my reasoning

145116146

The investigation ends quickly if I gather data that upholds my conclusion. So I put the three most uncomfortable questions next to the conclusion.

147117148

• Is the change in the method of the last attack the same criminal's adaptation or the traces of different criminals?
150119151
• How exactly does the cycle of a stolen vehicle overlap with the date of the attack
152120153
• Does the survivor's mask description refer to the same person?

155122156

These questions are not decorative. It is a switch that will change its conclusion when new data comes in. The ranking of the hypotheses changes immediately when directly linked information, such as the sighting source, preserved physical evidence, comparative samples from the same author, and independent dating. On the contrary, when the word "likely" increased, the ranking did not change.

Figure 8. FBI Open Investigation File Bundle — Image:

Final Reasoning: I Reached

159125160

Judgment
It is likely that one central criminal was involved in several attacks, and Eur Sweeney is a suspect with great explanatory power. However, beyond the spouse's statements and circumstances, the case cannot be closed under his name because there is no material evidence verified in the trial.

161126162

This conclusion is not a complete final judgment. The direction calls for the fewest assumptions in the currently disclosed records and explains the most clues at the same time. It is also clear that what comes out changes the conclusion. Reliable original records, verifiable legal science data, and independent evidence directly linking people and objects are needed.

163127164

I do not write that the case is closed. Instead, write down how far you can close it. That boundary is the result of this investigation. The fun of mystery is not that anything is possible. It's about finding out what you can throw away among the possible explanations.

Figure 9. Map connecting your attack points — Image:

Additional Survey Record 1. Redraw the timetable

167130168

Re-open the data on the desk and cover the conclusion this time. "Four attacks continued in the Texarkana area from February to May 1946."The place and the time were lined up, leaving only one behind. Unconfirmed sections were not filled by imagination but marked with blank squares. The larger the square, the lower the reliability of the hypothesis to pass through the section.

169131170

On top of it, the explanation was posted, "The same criminal launched a series of attacks by changing places and targets." At first, everything seemed right, but a weak connection was revealed when the question was asked, "Is the change in the way of the last attack the same criminal's adaptation or the trace of a different criminal." Hypothesis has to endure the question. If you keep making new coincidences to avoid questions, you will create, not reasoning, from that moment on.

171132172

The results obtained in this section are not confirmed criminals or complete plain texts. Instead, it became clear which explanations required fewer assumptions, and what would flip the ranking if found. Even if the case is not completed, the scope of the investigation can be narrowed. This record is intended to leave that very narrow boundary.

173133174

Further investigation record 2. Disappeared link

175134176

Re-expanded the data on the desk and covered the conclusion this time. "Five people were killed and three survived in the attack," he continued with a line of place and time. Unconfirmed sections were not filled by imagination but marked with blank squares. The larger the square, the lower the reliability of the hypothesis to pass through the section.

177135178

On top of it, the explanation was posted, "The first three cases are one criminal, and the last farm attack is an imitation or separate crime." At first, everything seemed right, but when I substituted the question, "How exactly does the cycle of stolen vehicles overlap with the date of the attack," the weak connection was revealed. Hypothesis has to endure the question. If you keep making new coincidences to avoid questions, you will create, not reasoning, from that moment on.

179136180

The results obtained in this section are not confirmed criminals or complete plain texts. Instead, it became clear which explanations required fewer assumptions, and what would flip the ranking if found. Even if the case is not completed, the scope of the investigation can be narrowed. This record is intended to leave that very narrow boundary.

181137182

Further investigation record 3. The path of movement of objects

183138184

Re-expanded the data on the desk and covered the conclusion this time. "Same ballistic features of multiple events.It pointed to a .32 caliber automatic pistol series."The place and the time were lined up, leaving only one behind. Unconfirmed sections were not filled by imagination but marked with blank squares. The larger the square, the lower the reliability of the hypothesis to pass through the section.

185139186

On top of it, I posted an explanation that "a local criminal using a stolen vehicle serialized an accidental attack." At first, everything seemed right, but when I substituted the question, "Does the description of the survivor's mask refer to the same person?" a weak connection was revealed. Hypothesis has to endure the question. If you keep making new coincidences to avoid questions, you will create, not reasoning, from that moment on.

187140188

The results obtained in this section are not confirmed criminals or complete plain texts. Instead, it became clear which explanations required fewer assumptions, and what would flip the ranking if found. Even if the case is not completed, the scope of the investigation can be narrowed. This record is intended to leave that very narrow boundary.

189141190

Further Survey Record 4. Opposite Hypothesis Night

191142192

I re-opened the material on the desk and covered the conclusion this time. "The site of the attack spanned roads to and from the state border and outlying farmhouses."The place and the time were lined up, leaving only one behind. Unconfirmed sections were not filled by imagination but marked with blank squares. The larger the square, the lower the reliability of the hypothesis to pass through the section.

193143194

On top of it, the explanation was posted, "The same criminal launched a series of attacks by changing places and targets." At first, everything seemed right, but a weak connection was revealed when the question was asked, "Is the change in the way of the last attack the same criminal's adaptation or the trace of a different criminal." Hypothesis has to endure the question. If you keep making new coincidences to avoid questions, you will create, not reasoning, from that moment on.

195144196

The results obtained in this section are not confirmed criminals or complete plain texts. Instead, it became clear which explanations required fewer assumptions, and what would flip the ranking if found. Even if the case is not completed, the scope of the investigation can be narrowed. This record is intended to leave that very narrow boundary.

197145198

Further Survey Record 5. Distance between record and memory

199146200

I re-opened the material on the desk and covered the conclusion this time. "Url Sweeney is on the main suspect line, but he has not been tried for murder," he continued in a line of place and time. Unconfirmed sections were not filled by imagination but marked with blank squares. The larger the square, the lower the reliability of the hypothesis to pass through the section.

201147202

On top of it, the explanation was posted, "The first three cases are one criminal, and the last farm attack is an imitation or separate crime." At first, everything seemed right, but when I substituted the question, "How exactly does the cycle of stolen vehicles overlap with the date of the attack," the weak connection was revealed. Hypothesis has to endure the question. If you keep making new coincidences to avoid questions, you will create, not reasoning, from that moment on.

203148204

The results obtained in this section are not confirmed criminals or complete plain texts. Instead, it became clear which explanations required fewer assumptions, and what would flip the ranking if found. Even if the case is not completed, the scope of the investigation can be narrowed. This record is intended to leave that very narrow boundary.

205149206

Additional Survey Record 6. Range of Last Scene

207150208

Re-open the data on the desk and cover the conclusion this time. "Four attacks continued in the Texarkana area from February to May 1946."The place and the time were lined up, leaving only one behind. Unconfirmed sections were not filled by imagination but marked with blank squares. The larger the square, the lower the reliability of the hypothesis to pass through the section.

209151210

On top of it, I posted an explanation that "a local criminal using a stolen vehicle serialized an accidental attack." At first, everything seemed right, but when I substituted the question, "Does the description of the survivor's mask refer to the same person?" a weak connection was revealed. Hypothesis has to endure the question. If you keep making new coincidences to avoid questions, you will create, not reasoning, from that moment on.

211152212

The results obtained in this section are not confirmed criminals or complete plain texts. Instead, it became clear which explanations required fewer assumptions, and what would flip the ranking if found. Even if the case is not completed, the scope of the investigation can be narrowed. This record is intended to leave that very narrow boundary.

Figure 10. Timeline that did not stop from February to May — Image:

Records and references

· FBI Vault — Texarkana Phantom Moonlight Murders

216156217

· Encyclopedia of Arkansas — Texarkana Moonlight Murders

218157219

※ This article is a commentary in the form of personal survey records based on public records and reliable data. I have distinguished between confirmed facts and my reasoning, and I update my judgment when a new official announcement is made. External material images were edited for some color and composition according to the permitted licensing conditions, and each image source was shown in the caption.