by Gordon Franz
This article is dedicated to my Maltese and American friends searching for the Apostle Paul’s shipwreck on Malta.
St. Paul’s Day – February 10, 2013
Have you ever watched a news broadcast where the meteorologist says that the next day there would be clear blue skies and it would be sunny all day? The presenter shows the radar screen, the forecast, and boasts how accurate their equipment is, so you plan a picnic at your favorite park for that day. Halfway through the picnic, however, the weather turns nasty with thunder and lightning and a torrential downpour! Forecasting weather is very unpredictable, more an art than science, even with sophisticated equipment.
Robert Cornuke presents a weather-related computer model of Paul’s shipwreck on Malta in his book, The Lost Shipwreck of Paul (2003: 184-193). I offer this objective critique of this model because of the serious nature of the issues involved.
During a Parliamentary debate on Malta in 2005, the Honorable Gavin Gulia asked the Prime Minister of Malta a Public Question (PQ 14720) about an affidavit that was sent to the United States Federal District Court in the state of Colorado for a trial between the former US Ambassador to Malta, Kathryn Proffitt, and Robert Cornuke. The reply to the Public Question states that:
“[The] Honourable Prime Minister Lawrence Gonzi said that he is informed that the affidavit was sent to safe-guard the reputation of the Armed Forces of Malta and of its officers because these had been misquoted in Bob Cornuke’s publication.“ (emphasis and highlight mine).
Since the issue has required the involvement of the government of Malta, let me add some additional analysis to the discussion that I hope will be helpful to interested parties.
The Computer Model on Malta
On Robert Cornuke’s third trip to Malta he gained access to “a very expensive and sophisticated computer program” at the Rescue Coordination Center of the Armed Forces of Malta on May 29, 2002. It was his hope that the data from this specialized computer model would “objectively speak to us across the millennia and trace the, until now, uncertain path of the biblical event of Paul’s journey from Crete to Malta” (2003: 184, plates 14-15; cf. Acts 27:8-28:1).
After the computer model was run on the hypothetical Alexandrian grain ship that carried the Apostle Paul and Dr. Luke, the course was shown approaching Malta more from the southeast, rather than directly from the east, the normal approach from Crete. The ship’s path line on the computer screen then intersected the East side of Malta, supposedly at the Munxar Reef and St. Thomas’ Bay preferred by Cornuke, not the traditional site for Paul’s shipwreck on the North side of the main Malta island, in the St. Paul’s Bay area. The model, it seemed, had overthrown tradition.
But Cornuke claimed the computer supported the Bible because Major Manuel Mallia, the Maltese officer in charge of the model, had agreed “that only St. Thomas Bay possessed all the physical, nautical, and geographical conditions that aligned perfectly with the Bible’s description [of Paul’s shipwreck]” (Cornuke 2003: 192-193; bracketed material and emphasis mine). Was this one of the misquotations by Cornuke in his book that required the involvement of the Maltese government? If this computer model is correct, however, it would help confirm Cornuke’s idea that the traditional location of Paul’s shipwreck was wrong.
Biblical Conflict with Cornuke’s Ideas
But there is a problem and it is a bigger one than tradition, it is a Biblical conflict: The East side of Malta, with the Munxar Reef and St. Thomas’ Bay, was familiar to Alexandrian ship captains as the side of the island they always saw on approach to Malta coming from the east. If sea captains could not make it back to Rome before the sea-lanes closed for the winter, the Alexandrian grain ships would dock in the Marsa Port on Malta (within the Grand Harbor of Valletta), off-load the grain and store it in granaries for the winter (Gambin 2005: 122-132; cf. Acts 28:11).
In Paul’s case they shipwrecked on a part of the island the crew did “not recognize” in fact so unfamiliar they did not even know they were on Malta until a native told them so (Acts 27:39; 28:1-2a, NKJV, emphasis added). So they had to have landed on some other part of Malta, not the familiar and recognizable East side. The traditional location on the unfamiliar North side of Malta makes sense in light of the puzzlement of the sailors on Paul’s wrecked ship. Computer or not, shipwrecking on the familiar East side makes no sense. This has always been a fundamental Biblical and logical stumbling block for Cornuke’s theory of Paul’s shipwreck because the seamen would have recognized the Munxar Reef and St. Thomas Bay, contrary to the Biblical text which states they did not recognize the island (Acts 27:39)! A computer model cannot overcome this fatal defect without simply throwing the whole Biblical account overboard in the process.
Two Principles of Computer Modeling
Even if we set aside the contradiction to the Biblical account for the moment, there are still major problems with the computer model and Cornuke’s use of it.
Two principles are important here:
- The computer model’s output will only be as good as the data inputted. There is a widely known axiom in the computer world, “Garbage In = Garbage Out,” which simply means that the computer results that come out are only as good as the data put in. If bad or mistaken data are put in, then the results will be bad or mistaken.
- Using a computer model beyond its design limitations and for purposes not intended will not produce trustworthy results. This could result in totally spurious results or results that can be easily manipulated to say almost anything, even unintentionally.
A computer model designed to assist search and rescue missions in the recent past hours or days of a modern-day storm causing a ship to go astray in AD 2002 (the year the model was run for Cornuke) obviously is not designed to reconstruct historical events from some 2,000 years ago – when there were no meteorological data from satellites and scientific instruments to plot shifting winds and currents. Even the ocean bottom can, and has, changed in two thousand years due to earthquakes and deposited silt.
The first step in understanding the design limitations of a computer model is to find out what model it is and what instructions it has for inputting data. Even better would be to have the developers’ design statement. Unfortunately, Cornuke did not even identify what computer model was used by the United States Coast Guard and the Maltese military!
Was the program purposely designed to recreate an actual past historical event and if so, what weather data were used for input? Cornuke does not provide the specific weather data inputs nor does he inform us where the input data came from. It would be impossible to know, for example, exactly what time the ship left Fair Haven on Crete, or precisely when and where the ship got caught in the Euroclydon (Northeaster storm) on its way to Phoenix on Crete (Acts 27:14) because the Bible does not state this information. Did the storm strike as soon as they left Fair Haven, or several hours later, right before they were to dock at Phoenix? Or, was it somewhere in between the two places? Each of these unknown variables would affect the geographic location of where the ship ended up in the output of the computer model.
Data Input for the Computer Model
According to Cornuke, the modelers used five types of input data for the model (Cornuke 2003: 187-188). These included:
1. The “general parameters of a grain freighter”
One nautical archaeologist has pointed out, however, that “the precise appearance of great grain ships like those mentioned in the Book of Acts and the writings of Lucian” are unknown (Fitzgerald 1990: 31) because nautical archaeologists have never recovered an actual first century AD Alexandrian grain ship in an underwater archaeological excavation. Was the grain ship a two-mast or a three-mast ship? What was its draft? How much did it actually weigh? Cornuke said they put in “the approximate size of the ship” (2003: 187, emphasis mine), yet a variation in size and weight would affect the outcome of the calculations for the computer model.
2. Wooden hull was a factor entered into the software
But was only wood exposed on the hull of Paul’s ship or was there lead sheathing on the hull? Ancient lead sheathing has been found on the seabed of Malta. If there was lead sheathing on the grain ship that would affect the outcome of the calculations.
Also, the ship was undergirded, probably with heavy rope or cable (Hirschfeld 1990: 26-27), to secure it during the storm (Acts 27:17). What effect would the rope or cable have on the drag of the ship and thus on the computer calculations?
3. The “veering characteristics of a northeaster”
Cornuke suggested the drag of the windsock affected the speed and direction of the ship (Cornuke 2003: 190). What ancient sources describe – or archaeological remains show – that a windsock sail was part of a rigging for an Alexandrian grain ship and used as a sea anchor in an emergency? I am not aware of any. Perhaps Cornuke can enlighten us with this information.
4. The “leeway of time”
What margin of error or maximum variation (leeway) in the “time” is meant – and is it maximum variation in the time of day or the time of year? It is unclear. How was the possible variation of time factored in? Did they run the computer with every possible choice of time? What were the results?
Cornuke had the rescue software run on May 29, 2002. The question is then: Did they run the software model with the current date of May 29, or did they think to change the date to the Fall season? (Shipwreck occurred at least 14 days after Yom Kippur and before winter, thus most probably October-November: cp. Acts 27:9, 27, 33; 28:11.) In fact, does the computer model even differentiate a year as well as the day of the year, and if so, was the year AD 2002 run or a year around ca. AD 60 when Paul’s ship wrecked?
The ocean currents in the Fall were programmed into the computer model (see item 5 below) but it is unclear whether a Fall date was also entered for wind speeds and directions. If they did change the computer model date to the Fall, what date in the Fall did they choose? There is no explanation given to clarify any of this.
5. The currents during the Fall season for that part of the Mediterranean Sea
Although Cornuke listed five types of data inputted into the computer model including ocean currents, he strangely failed to list winds even though powerful storm winds are far more important than ocean currents. Wind directions and speeds are the critical factors in a storm of this apparent magnitude. The exact wind speeds and directions are unknown and any increase or decrease in speed, or change of wind direction, from hour to hour and day to day, would affect the outcome of the computer model over the 14 days the grain ship was adrift.
Unfortunately, the specific information that was put into the computer was not given in the book, perhaps because it is a popular-level book. But the specific input data were not provided on Cornuke’s websites or in any peer-reviewed scholarly article either (none have been published). Researchers who would like to follow up or try to duplicate this computer exercise would need the specific information inputted into the computer software, such as the wind speeds and directions and ocean currents hour by hour, what alternative dates, times, winds and currents were used and with what results, etc.
The Computer Model’s “Line of Drift”
Plates 14 and 15 of The Lost Shipwreck of Paul display photographs of the computer experiment at the Rescue Coordination Center of the Armed Forces of Malta. On the bottom of Plate 14, the line of drift for Paul’s Alexandrian grain ship is drawn. I enlarged the photograph on a photocopy machine to 200% and examined the “line of drift.” It appears to be drawn by human hand with a felt tip pen or magic marker, not by computer. The thickness of the line seems to vary slightly and at one point the line seems to be redrawn over a short segment where it is a bit thicker. At another point the line does not have an even, smooth flow to it. This seems to suggest that the line is hand-drawn and not computer generated.
I also observed that the line of drift was not drawn through the last datum point but rather above it. Why was this? This last datum point also seems to fall far short of reaching the Malta area so it would be interesting to know, if the computer had generated one more datum point, just where that last point would have been located. When I redrew the line (see chart below) through the last datum point that is shown, and not above it as represented in the book, the line of drift misses Malta entirely, by about 5 miles to the south of the island! Thus it does not hit the Munxar Reef and St. Thomas’ Bay as Cornuke claims.
Technical, Peer-Reviewed Article is Needed
For Cornuke’s research to be evaluated by scholars, it must be published in a peer-reviewed scientific publication, perhaps a meteorological journal, identifying the software program that was used and the specific input information used to simulate the storm. An explanation is also in order as to why the “line of drift” did not go through the last datum point and if there was one more computer-generated datum point. If there was another datum point, where was it?
If independent researchers could replicate his research using the same or similar software then Cornuke’s research would have added credibility and congratulations would be in order. Or, perhaps with slight variations in the different variables, the computer model might have the grain ship run aground in the traditional St. Paul’s Bay area or completely miss the island, as presently appears to be the case!
It would also be helpful if Cornuke could have Major (now Colonel) Manuel Mallia of the Rescue Coordination Center, who ran the model for Cornuke, provide a letter indicating the model’s appropriateness for the task, some of the key data input, and stating whether he agreed or disagreed with the conclusions Cornuke drew from the output.
The Conclusion of the Matter
A word of caution is in order. Computer models are great tools for predicting the outcome of various data sets entered into the model. But while they are excellent modeling tools, they are simply that – tools to generate possible outcomes. They are seldom the final word on what will certainly happen in the future, and for sure, not the final word on what did happen in the unknown past.
The weather prediction by the meteorologist is based on a large volume of recently obtained weather data from instruments put into a computer model. The U.S. National Oceanic and Atmospheric Administration (NOAA) explains the procedure on its official website:
The [computer weather-prediction] models, using many millions of numbers that represent weather [observation] parameters such as temperature, pressure, wind, etc., attempt to represent current weather conditions and then make a prediction of the future state of the atmosphere….
Data Assimilation is the process whereby weather observations are incorporated into a computer model that predicts the weather. After billions of calculations, the supercomputers that are now used to run weather models, project how the current weather conditions are expected to change.
But for Paul’s shipwreck, we have zero weather instrument data, there were no weather instruments in that era and only fragmentary records of human events and occasional weather events such as major once-a-century type storms. Any “data” is invented by extrapolating current conditions and data back in time 2,000 years and assuming that past weather was exactly the same as today. Needless to say this is highly speculative at best and non-verifiable.
Consider how many computer weather models have predicted hurricanes that never materialized or missed significant weather events that actually took place. How often have you noticed that your local weather forecast has been right?! (Or wrong, and it ruined your picnic!). Today’s weather forecasts attempt to project a few hours or days into the future. In this scenario, a meteorologist’s forecast has everything in its favor, yet sometimes it is still incorrect. By contrast, a computer model of the possible location of Paul’s shipwreck attempts to project conditions back nearly 2,000 years into the past. It is far from definitive given so many unknown variables and factors. Thus, we should not put too much stock in such fantastic extrapolations!
Also, depending on the input, the same model could have easily produced a completely different location for the shipwreck, including even the traditional location of the St. Paul’s Bay area. Perhaps the most difficult data to input for this, or any model, is the sovereign Hand of God controlling the speed and direction of the wind and thus, the precise, final destination of the Alexandrian grain ship!
For links to other critiques of Cornuke’s ideas, see:
2003 The Lost Shipwreck of Paul. Bend, OR: Global Publishing Services.
1990 The Ship of Saint Paul. Comparative Archaeology. Biblical Archaeologist 53/1: 31-39.
2005 The Maritime Landscapes of Malta from the Roman Period to the Middle Ages. Unpublished Ph.D. Dissertation University of Bristol [England].
1990 The Ship of Saint Paul. Historical Background. Biblical Archaeologist 53/1: 25-30.
About the author
Gordon Franz is a Bible teacher who holds an MA in Biblical Studies from Columbia Biblical Seminary, SC. Since 1978, he has engaged in extensive research in Biblical geography and archaeology and has participated in a number of excavations in and around Jerusalem, including Ketef Hinnom and Ramat Rachel as well as the excavations at Lachish, Jezreel, Hazor, and Tel Zayit. He has taught the geography of the Bible and led field trips in Israel for the Jerusalem Center for Biblical Studies, the Institute of Holy Land Studies, and the IBEX program of The Master’s College. He also co-teaches the Talbot School of Theology’s Bible Lands Program. He has also visited Malta on a number of occasions since January 1997 doing research on the history, geography, and archaeology of the island, as well as the location of Paul’s shipwreck. Gordon is on the staff of the Associates for Biblical Research.