Estimating trajectories of colonisation to the Mariana Islands, western Pacific.

Fitzpatrick, Scott M. ; Callaghan, Richard T.

[ILLUSTRATION OMITTED]

Introduction

The Marianas island chain, located in the north-western tropical Pacific, is one of the most remote island groups in the western Pacific Basin (Figure 1). The archipelago stretches for over 800km in a north-south configuration between 13[degrees] and 20[degrees] N. It is 1800km from New Guinea and the Bismarcks to the south, and up to 2300km away from the northern Philippines and Taiwan to the west. Current archaeological research suggests that the Marianas island chain was settled by at least 3200 cal BP, and possibly as early as 3500 cal BP (see Carson 2008, 2011; Clark et al. 2010; Hung et al. 2011; Carson & Kurashina 2012). Early radiocarbon dates from Ritidian on Guam and Unai Bapot on Saipan (Carson 2008, 2011; Clark et al. 2010; Hung et al. 2011) are associated with thin, calcareous sand Tempered (CST), red-slipped pottery known as Marianas Red Ware (first described by Spoehr 1957). Similar pottery has been found at other sites, including Achugao on Saipan (Butler 1994) and the House of Taga site on Tinian (Pellett & Spoehr 1961), though the latter lacks rigorous dating that could chronologically anchor the site and associated artefacts.

[FIGURE 1 OMITTED]

This chronology is contemporaneous with, or even slightly earlier than, that of Lapita groups who began moving eastward from Melanesia into West Polynesia c. 3400-3300 BP, an observation that makes this an astonishing feat in the history of Pacific island colonisation. Not only do the Marianas lie between 1800 and 2300km from the nearest mainland source--thought to be somewhere in Island Southeast Asia (ISEA)--but there are very few intervening islands to use as stepping stones, regardless of which mainland source was the original departure point. That would give the Marianas chain the distinction of being the most remote, and earliest, island chain to have been settled in Remote Oceania, involving a sea crossing of some 2000km. This is in stark contrast to Lapita peoples who initially crossed gaps of only hundreds of kilometres.

While the Marianas may represent the longest sea crossing known in Pacific island prehistory, the origin of these initial colonisers is still debated. Hung et al. (2011), for example, have postulated that the first settlers originated from the northern Philippines. That is based on typological similarities and the chronological overlap of Marianas Red Ware with red-slipped, incised and stamped pottery found at the Nagsabaran site in the Cagayan Valley of north-eastern Luzon in the Philippines, which dates to 2000-1000 BC. There are also linguistic and genetic affiliations.

In a rejoinder to Hung et al, (2011), Winter et al. (2012) instead proposed that the ceramics, linguistics and oceanographic conditions point to an origin further south in ISEA. They note that it would have been extremely difficult, if not impossible, to drift or sail eastward from the Philippines to the Marianas given prevailing currents and winds and without an adequate sailing technology like the double-hulled canoe with fixed mast and standing rigging. Instead, they note that the area between Mindanao in the southern Philippines and the Bird's Head of New Guinea serves as a more suitable corridor for venturing into western Micronesia given the fairly consistent south-west winds, particularly between July and September.

Winter et al. (2012) also conducted analysis of ceramics from Nagsabaran and Unai Bapot and noted a number of important technological differences in pottery manufacture, including the dominance of calcareous sand temper (CST) in the Marianas assemblages and the use of paddles and anvils, versus the natural volcanic sand temper and coiling that were preferred at Nagsabaran. It should be noted, however, that the sample size was fairly small (eight vessels from the Philippines and nine from the Unai Bapot site in the Marianas; Winter et al. 2012). Furthermore, additional analysis of the Marianas pottery manufacturing sequence (M. Carson pers. comm.) has identified a combination of slab-built and coil-built primary forming, followed by paddle-anvil and trimming as a secondary stage. Finally, Winter et al. (2012) argue from the corpus of linguistic evidence that Chamorro (the indigenous language of the Marianas) does not seem to have had a northern Philippine origin, but instead to have dispersed from ISEA where Malayo-Polynesian was spoken, though Blust (2000) has noted a number of similarities between Chamorro and languages found in the central Philippines (or perhaps elsewhere in the archipelago).

While additional archaeological, linguistic and genetic data are needed to resolve the origin of early Marianas colonists, one important avenue of research that has not yet been fully addressed, and which can complement other lines of evidence, is the use of computer simulations of seafaring. The issue hinges on whether the islands could have been settled purposefully versus accidentally given the prevailing oceanographic conditions (winds, currents). We approach this problem using a computer simulation program to estimate the likelihood that prehistoric seafarers reached Guam or other islands in the Marianas chain from different points around the western Pacific. These simulations have been used by various researchers around the world for both prehistoric and historic voyages to estimate the probability that seafarers (downwind/drift or intentional) leaving one location would teach another, the trajectories they would take, and the likelihood of crew survival (e.g. Levison et al. 1973; Irwin 1992, 2010; Di Piazza & Pearthree 1999; Montenegro et al. 2006, 2008; Avis et al. 2007; Di Piazza et al. 2007). In this paper, we use the seafaring simulation program known as Seascape, which was developed by one of the authors (R.C.) and has been used in numerous other successful projects (e.g. Callaghan 2001, 2003a, b & c, 2008; Callaghan & Bray 2007; Callaghan & Fitzpatrick 2007, 2008; Fitzpatrick & Callaghan 2008a & b, 2009). Our goal is not to review the archaeological, linguistic or genetic data to decide whether one part of Island Southeast Asia or Melanesia was a more likely origin for Chamorro groups. We seek instead to provide additional data on seafaring probabilities to estimate the most likely trajectory, and thereby to complement other lines of evidence.

Below we detail the functions and parameters of the Seascape simulation program and discuss the findings of both drifting (downwind) and intentional (directed) voyaging scenarios for each month of the year for Taiwan, Luzon, Mindanao, Halmahera, western New Guinea, central New Guinea, eastern New Guinea and the Bismarck Archipelago. These represent the most likely points of departure for colonists to the Marianas. We also simulated directed voyages from Guam to the Philippines. The results suggest that there is 0% probability during any time of the year of either downwind or directed voyages being able to colonise the Marianas from Taiwan or the Philippines. In general, the highest success rates for directed voyages were from Halmahera, the Bismarcks or points along the northern New Guinea coastline. Overall, the only successful downwind or directed voyages from any location we tested were those undertaken between July and October, indicating that there was a narrow window of time during the year when colonisation is likely to have occurred.

Methods

The Seascape simulation program uses wind and current data for each month of the year from the CD-ROM version of the U.S. Navy's Marine climatic atlas of the world (Version 1.1; U.S. Navy 1995) (see Callaghan 2001, 2003a, b & c, 2008; Callaghan & Bray 2007; Callaghan & Fitzpatrick 2007, 2008; Fitzpatrick & Callaghan 2008a & b, 2009; Callaghan & Scarre 2009). With the exception of Arctic regions, the data in the Atlas (U.S. Navy 1995) include all of the world's oceans and seas.

The actual mechanics of the program (how the simulations are structured) includes ways in which the data are organised and selected, how success is calculated as a percentage of the total of voyages, and a number of variables. Four main variables are considered when conducting these seafaring simulations: 1) vessel type; 2) propulsion; 3) wind patterns; and 4) current patterns.

The data are organised in a resolution of one-degree Marsden squares (one degree of latitude x one degree of longitude). The program randomly selects wind and current data that are frequency-weighted according to the compiled observations of the Marine Climatic Atlas. These forces are then allowed to operate on vessels for a 24-hour period before a new selection is made (see Levison et al. 1973 for a justification of the period length). Success is simply the percentage of voyages reaching a selected target within a survivable time.

Vessel type and propulsion are important variables of any simulation, as the shape of an object both above and below the waterline affects its response to the wind. Polar diagrams can be used to determine performance characteristics (e.g. Di Piazza et al. 2007: 1221). These diagrams, which give the ratio of vessel velocity to true wind velocity, are available for a number of Polynesian sailing vessel types. Performance characteristics can also be determined by held experimentation (e.g. Gladwin 1970; Finney 1979, 2003; Lewis 1994) which is particularly useful for non-sailing craft.

Objects caught within a current are affected by a 1:1 ratio (i.e. unless other forces are also operating, the object has the same set and speed as the current). Unless the current is exceptionally strong, any object floating with an appreciable part above the waterline will be more affected by wind than by currents. Traditional navigators in Kiribati, for example, used the effect of wind on objects floating high in the water to determine the direction of land when recent winds differed from the flow of the current (Lewis 1994).

The actual distance and direction travelled are based on the wind and current data combined with the speed of vessels in relation to the wind (from Levison et al. 1973) and parameters selected by the program operator. Parameters of the simulation are choices made by the program operator in order to set up the simulation to answer a particular question. This includes the following information: 1) point of origin and destination; 2) crew strategy; 3) performance characteristics; 4) duration of voyages; 5) time of year; and 6) number of simulations.

In this study, multiple points of origin were used: Taiwan; Luzon; Mindanao; Halmahera; western New Guinea; central New Guinea; eastern New Guinea; and the Bismarck Archipelago. We also simulated directed voyages from Guam to the Philippines. Vessels were positioned as close to the centres of Marsden squares as possible, but in some cases were moved further away, as vessels cannot be started on land. This was an issue discovered by Levison et al. (1973), who noted that vessels starting too close to shore tended to return to the same coast.

Two crew strategies were simulated. The first was downwind sailing, where the crew simply tries to cover as much distance as possible. This is a strategy often used when the vessel's position has been lost due to storms (Dening 1963: 138-53). The second strategy was directed sailing where the crew selects a heading. Here, no intent was assumed for the crew other than to travel in the direction of the Marianas or, in the case of Guam, to the Philippines due west.

In this instance, we do not know what types of vessels were used, although some form of outrigger canoe is most likely. Given the overall uncertainty of vessel type, we have chosen to follow Levison et al. (1973: 19, 27) and use more generalised figures that are reasonable for a variety of watercraft types. Levison et al. (1973: 27), basing their estimation for windward sailing on experiments by Horvath and Finney (1969) and Bechtol (1963: 100), and observations of modern canoes by Gladwin (1970), assumed that vessels could only sail on a teach (perpendicular to the wind) at 90[degrees]. This is more conservative than actual experiments and observations suggest. It is also assumed that when winds are on the beam or aft, the vessels could maintain speeds equivalent to those adapted for downwind sailing.

The duration of voyages is a critical factor in crew survival. Some indication of realistic maximum durations for voyages can be derived from lengthy drift voyages in open boats, due, for example, to shipwreck or other misfortune. These are well known in the Pacific both historically and in modern times (Howay 1944; Levison et al. 1973: 20-21). The maximum drift voyage seems to be of the order of seven to eight months. Several voyages recorded covered distances of nearly 5000km over a period of six to ten weeks, and a great number covered shorter distances. Given that intentional voyages with the supplies and equipment necessary for long durations at sea (rather than drift voyages) are also considered here, a maximum duration of 200 days was set for the simulations.

Time of year is important, as in most regions of the world there is significant seasonal variation in wind and current patterns. Vessels were started from all points of origin for all 12 months of the year. The program automatically shifts to the database for the following month after the month originally selected has expired, reflecting the changing wind and current conditions.

The number of simulations from each origin point is 100 for each month. Earlier work (Callaghan 1999) conducted 1000 simulations for each origin point. Only very rarely, however, did an outlier occur different from those given by the smaller sample of 100, prompting the use of a sample size of 100 in subsequent analyses.

Results

Out simulations demonstrated that for downwind sailing (Table 1), there was no possibility of reaching the Marianas at any time of the year from Taiwan, Luzon or Mindanao. Overall success is highest from central New Guinea, with 13% of vessels landing in the Marianas if starting in August, 25% in September (Figure 2) and 2% in October Durations of voyages ranged from 58 to 198 days. Success rates were the second highest from the Bismarck Archipelago, with 1% of vessels starting in July reaching the Marianas, 4% in August, 13% in September (Figure 3) and 9% in October. Durations of voyages ranged from 54 to 191 days. Vessels starting from western New Guinea had the next highest success rate. Vessels starting in July and August had 3% success in reaching the Marianas, 7% in September and 8% in October. Durations of voyages ranged from 56 to 184 days. From eastern New Guinea, success rates were 6% in August and September and 3% in October. Durations of voyages were from 84 to 192 days. Finally, from Halmahera, success was 9% in September and 2% in October. Durations were from 56 to 137 days.

As with downwind voyages, successful directed voyages were not possible from Taiwan, Luzon or Mindanao (Table 2). From the remaining origin points, it was possible to reach the Marianas from July to October, with the exception of central New Guinea where success was possible from August to October. Success rates did not follow the same pattern as for downwind voyages. Vessels sailing from the Bismarck Archipelago had the highest success, with 14% in July, 32% in August (Figure 4), 24% in September and 18% in October. Durations of voyages lasted from 30 to 98 days. From Halmahera, vessels had a 13% chance of reaching the Marianas in July, 27% in August (Figure 5), 20% in September and 10% in October. Durations of voyages were from 22 to 184 days. From western New Guinea, success rates were 5% in July, 10% in August, 24% in September and 26% in October (Figure 6). Voyages lasted from 37 to 144 days. From eastern New Guinea, voyages starting in July had a 9% success rate, 17% in August and September and 14% in October. Voyages lasted between 35 and 98 days. Interestingly, central New Guinea, which had the highest downwind sailing success rate, had the lowest rate for directed voyages. Those voyages beginning in August had only a 2% rate of success, 20% in September and 18% in October. Voyages lasted between 29 and 94 days.

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

Generally, if it is difficult to sail in a direction due to contrary wind and current patterns, it will be easier to sail in the opposite direction. This is the case for directed voyages from Guam to the west (Table 3). However, in July and August, vessels attempting to sail west towards the Philippines are all swept northwards by winds and currents. With a June departure, only one vessel reached the Philippines, while two reached Taiwan. The rest of the voyages were swept northwards by the Kuroshio Current. The results for September were similar, with only three vessels reaching the northern Philippines. The rest of the voyages went north to China and Taiwan. Between October and May, success rates for vessels reaching the Philippines were high and nearly all vessels reached land. In October conditions, 96% of vessels landed in the Philippines and the rest made landfall to the south. In November, 91% of vessels made landfall in the Philippines, with 8% landing on the New Guinea coast and 1% on Palau. Given a start with December conditions, 77% of the vessels reached the Philippines, while 23% made landfall to the south. For a January start, success was 73%, with 23% of vessels landing on Halmahera or the Moluccas, 2% on Palau, 1% in western New Guinea and 1% on Yap. With February starting conditions, 33% of vessels landed in the Philippines, 58% in Halmahera and the Moluccas, 3% in western New Guinea and 6% in Sulawesi. For March, 52% of vessels reached the Philippines and 44% reached Halmahera and the Moluccas, while 4% landed in New Guinea. With April starting conditions, 73% of voyages ended in the Philippines, while the rest moved north. Starting in May, 65% of vessels reached the Philippines and the remaining voyages terminated in Taiwan and China.

[FIGURE 4 OMITTED]

An interesting observation is that at some times of the year, there is a high probability that vessels sailing downwind from all of the origin points other than Guam will end up in the Philippines. From Halmahera in October, 20% of downwind voyages land in the Philippines. From western New Guinea in September, 63% make the same landfall. For central New Guinea, 63% of voyages land in the Philippines starting in October. Starting in September, 26% of voyages make a Philippine landfall, and with October start conditions, 34% of vessels from the Bismarcks land in the Philippines.

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

Discussion and conclusions

The foregoing simulations allow us to evaluate the competing models for the colonisation of the Marianas from the northern Philippines (Hung et al. 2011) or from further south (Winter et al. 2012). Our results indicate that there is a 0% chance of either intentional (directed) or unintentional (drift) voyaging to the Marianas from Taiwan or anywhere in the northern Philippines (Tables 1 & 2). This casts doubt on the model advocated by Hung et al. (2011). There is, however, a reasonable chance of reaching the Marianas from Halmahera, the Bismarcks or anywhere along the northern New Guinea littoral, especially by directed voyages. The success rate is as high as 32% for intentional voyaging to Guam from the Bismarcks and 27% from Halmahera during the month of August (Table 2).

Interestingly, while it is impossible to reach Guam from Taiwan or the northern Philippines under prevailing oceanographic conditions, vessels leaving Guam generally have a high rate of success in reaching the Philippines during the late autumn and early winter months (see Table 1). It is theoretically possible that anomalous conditions such as a storm event could have driven initial colonisers to the Marianas from the northern Philippines--at which point they could much more easily have voyaged back (intentionally or not) to their supposed homeland during certain times of the year. This scenario appears unlikely, however, from the simulation evidence, particularly since directed voyages from the northern Philippines were never successful at any time of year.

It is difficult to say with any degree of certainty whether the initial voyages were undertaken by accident or design. Population movements in other parts of ISEA testify to increasingly sophisticated watercraft and navigational skills during the Late Holocene, however, and it is not outside the realm of possibility that people consistently pursued new islands to colonise, albeit with a low rate of initial success. The further possibility exists that some islands were 'discovered' but not fully colonised until later in time.

The early radiocarbon dates from the Marianas, dating to as early as c. 3500 BP, are indicative of a lengthy voyage from some point in ISEA or the northern New Guinea region, even when intervening islands are considered. From many lines of evidence, it appears that communication between the Philippines and islands in western Micronesia occurred periodically, though at different scales and levels of intensity through time. While pottery similar to Marianas Red Ware has not been found in Indonesia to date, the simulations we conducted emphasise that additional linguistic, genetic and archaeological research is needed to resolve the question of Chamorro origins, with a departure point further south being a distinct possibility.

References

AVIS, C., A. MONTENEGRO & A. WEAVER. 2007. The discovery of western Oceania: a new perspective. Journal of Island and Coastal Archaeology 2: 197-209.

BECHTOL, C. 1963. Sailing characteristics of oceanic canoes, in J. Golson (ed.) Polynesian navigation (Polynesian Society Memoir 34): 98-101. Auckland: Polynesian Society.

BLUST, R.A. 2000. Chamorro historical phonology. Oceanic Linguistics 39: 83-122.

BUTLER, B.M. 1994. Early prehistoric settlement in the Mariana Islands: new evidence from Saipan. Man and Culture in Oceania 10: 15-38.

CALLAGHAN, R.T. 1999. Computer simulations of ancient voyaging. Northern Mariner 9: 11-22.

--2001. Ceramic age seafaring and interaction potential in the Antilles: a computer simulation. Current Anthropology 42: 308-13.

--2003a. Comments on the mainland origins of the Preceramic cultures of the Greater Antilles. Latin American Antiquity 14: 323-38.

--2003b. Prehistoric trade between Ecuador and west Mexico: a computer simulation of coastal voyages. Antiquity 77: 796-804.

--2003c. The use of simulation models to estimate frequency and location of Japanese Edo period wrecks along the Canadian Pacific Coast. Canadian Journal of Archaeology 27: 62-82.

--2008. On the question of the absence of Archaic Age sites on Jamaica. Journal of Island and Coastal Archaeology 3: 54-71.

CALLAGHAN, R.T & W. BRAY. 2007. Simulating prehistoric sea contacts between Costa Rica and Colombia. Journal of Island and Coastal Archaeology 2: 4-23.

CALLAGHAN, R.T. & S.M. FITZPATRICK. 2007. On the relative isolation of a Micronesian archipelago during the historic period: the Palau case study. International Journal of Nautical Archaeology 36: 353-64.

--2008. Examining prehistoric migration patterns in the Palauan archipelago: a computer simulated analysis of drift voyaging. Asian Perspectives 47: 28-44.

CALLAGHAN, R.T. & C. SCARRE. 2009. Simulating the western seaways. Oxford Journal of Archaeology 28: 357-72.

CARSON, M.T. 2008. Refining earliest settlements in Remote Oceania: renewed archaeological investigations at Unai Bapot, Saipan. Journal of Island and Coastal Archaeology 3: 115-39.

--2011. Paleohabitat of first settlement sites 1500-1000 BC in Guam, Mariana Islands, western Pacific. Journal of Archaeological Science 38: 2207-21.

CARSON, M.T. & H. KURASHINA. 2012. Re-envisioning long-distance oceanic migration: early dates in the Mariana Islands. World Archaeology 44: 409-35.

CLARK, G., F. PETCHEY, O. WINTER, M. CARSON & P. O'DAY. 2010. New radiocarbon dates from the Bapot-1 site in Saipan and Neolithic dispersal by stratified diffusion. Journal of Pacific Archaeology 1: 21-35.

DENING, CM. 1963. The geographical knowledge of the Polynesians and the nature of inter-island contact, in J. Golson (ed.) Polynesian navigation (Polynesian Society Memoir 34): 138-53. Auckland: Polynesian Society.

DI PIAZZA, A. & E. PEARTHREE. 1999. The spread of the 'Lapita people': a demographic simulation. Journal of Artificial Societies and Social Simulation 2(3). Available at: http://www.soc.surrey.ac.Uk/JASSS/2/3/4.html (accessed 16 April 2013).

DI PIAZZA, A., P. DI PIAZZA & E. PEARTHREE. 2007. Sailing virtual canoes across Oceania: revisiting island accessibility. Journal of Archaeological Science 344: 1219-25.

FINNEY, B. 1979. Hokule'a, the way to Tahiti. New York: Dodd, Mead & Co.

--2003. Sailing in the wake of the ancestors: reviving Polynesian voyaging. Honolulu: Bishop Museum.

FITZPATRICK, S.M. & R.T. CALLAGHAN. 2008a. Magellan's crossing of the Pacific: using computer simulations to examine how oceanographic conditions influenced one of the world's greatest voyages. Journal of Pacific History 48(2): 145-65.

--2008b. Seafaring simulations and the origin of prehistoric settlers to Madagascar, in G. Clark & S. O'Connor (ed.) Islands of inquiry: papers in honour of Atholl Anderson (Terra Australis 29): 55-66. Canberra: Australian National University Press.

--2009. Examining dispersal mechanisms for the translocation of chicken (Gallus gallus) from Polynesia to South America. Journal of Archaeological Science 36: 214-23.

GLADWIN, T. 1970. East is a big bird: navigation and logic on Puluwat Atoll. Cambridge (MA): Harvard University Press.

HORVATH, S.M. & B. FINNEY. 1969. Paddling experiments and the question of Polynesian voyaging. American Anthropologist 71: 271-76.

HOWAY, F.W. 1944. Some lengthy open-boat voyages in the Pacific Ocean. American Neptune 4: 53-57.

HUNG, H-C., M.T. Carson, P. Bellwood, F.Z. CAMPOS, P.J. PIPER, E. DIZON, M.J.L.A. BOLUNIA, M. OXENHAM & C. ZHANG. 2011. The first settlement of Remote Oceania: the Philippines to the Marianas. Antiquity 85: 909-26.

IRWIN, G.J. 1992. The prehistoric exploration and colonisation of the Pacific. Cambridge: Cambridge University Press.

--2010. Pacific voyaging and settlement: issues of biogeography and archaeology, canoe performance and computer simulation, in A. Anderson, J.H. Barrett & K.V. Boyle (ed.) The global origins and development of seafaring. 131-41. Cambridge: McDonald Institute for Archaeological Research.

LEVISON, M., R.G. WARD & J.W. WEBB. 1973. The settlement of Polynesia, a computer simulation. Canberra: The Australian National University.

LEWIS, D. 1994. We, the navigators. Honolulu: University Press of Hawai'i.

MONTENEGRO, A., R. HETHERINGTON, M. EBY & A.J. WEAVER. 2006. Modelling pre-historic transoceanic crossings into the Americas. Quaternary Science Reviews 25: 1323-38.

MONTENEGRO, A., C. AVIS & A.J. WEAVER. 2008. Modelling the pre-historic arrival of the sweet potato in Polynesia. Journal of Archaeological Science 35: 355-67.

PELLETT, M. & A. SPOEHR. 1961. Marianas archaeology: report on an excavation on Tinian. Journal of the Polynesia Society 70: 321-25.

SPOEHR, A. 1957. Marianas prehistory: archaeological survey and excavations on Saipan, Tinian and Rota (Fieldiana: Anthropology 48). Chicago (IL): Chicago Natural History Museum.

U.S. Navy. 1995. Marine climatic atlas of the world. Asheville (NC): National Climatic Data Center.

WINTER, O., G. CLARK, A. ANDERSON & A. LINDAHL. 2012. Austronesian sailing to the northern Marianas, a comment on Hung et al. (2011). Antiquity 86: 898-910.

Received: 24 July 2012; Accepted: 13 September 2012; Revised: 20 November 2012

Scott M. Fitzpatrick (1) & Richard T. Callaghan (2)

(1) Department of Anthropology, University of Oregon, Eugene, OR 97403, USA

(2) Department of Archaeology, University of Calgary, 2500 University Dr. NW, Alberta, T2N1N4, Canada

Table 1. Success rates for downwind voyages to the Mariana Islands
and duration in days (in parentheses).

Code Taiwan Luzon Mindanao Halmahera

January 0% 0% 0% 0%
February 0% 0% 0% 0%
March 0% 0% 0% 0%
April 0% 0% 0% 0%
May 0% 0% 0% 0%
June 0% 0% 0% 0%
July 0% 0% 0% 0%
August 0% 0% 0% 0%
September 0% 0% 0% 9% (56-137)
October 0% 0% 0% 2% (57-91)
November 0% 0% 0% 0%
December 0% 0% 0% 0%

Code W. New Guinea C. New Guinea

January 0% 0%
February 0% 0%
March 0% 0%
April 0% 0%
May 0% 0%
June 0% 0%
July 3% (77-181) 0%
August 3% (56-184) 13% (60-175)
September 7% (101-141) 25% (58-198)
October 8% (65-178) 2% (63-116)
November 0% 0%
December 0% 0%

Code E. New Guinea Bismarck Archipelago

January 0% 0%
February 0% 0%
March 0% 0%
April 0% 0%
May 0% 0%
June 0% 0%
July 0% 1% (90)
August 6% (84-192) 4% (137-176)
September 6% (106-138) 13% (54-158)
October 3% (166-175) 9% (76-191)
November 0% 0%
December 0% 0%

Table 2. Success rates for directed voyages to the Mariana Islands
and duration in days (in parentheses).

 Taiwan Luzon Mindanao Halmahera W. New Guinea

January 0% 0% 0% 0% 0%
February 0% 0% 0% 0% 0%
March 0% 0% 0% 0% 0%
April 0% 0% 0% 0% 0%
May 0% 0% 0% 0% 0%
June 0% 0% 0% 0% 0%
July 0% 0% 0% 13% (38-146) 5% (49-109)
August 0% 0% 0% 27% (22-184) 10% (37-71)
September 0% 0% 0% 20% (30-160) 24% (39-101)
October 0% 0% 0% 10% (39-160) 26% (42-144)
November 0% 0% 0% 0% 0%
December 0% 0% 0% 0% 0%

 C. New Guinea E. New Guinea Bismarck Archipelago

January 0% 0% 0%
February 0% 0% 0%
March 0% 0% 0%
April 0% 0% 0%
May 0% 0% 0%
June 0% 0% 0%
July 0% 9% (35-98) 14% (35-82)
August 2% (29-53) 17% (39-93) 32% (39-82)
September 20% (40-46) 17% (39-74) 24% (34-76)
October 18% (41-94) 14% (41-76) 18% (30-98)
November 0% 0% 0%
December 0% 0% 0%

Table 3. Success rates for directed voyages to the Philippines
from Guam.

January 73%
February 33%
March 52%
April 73%
May 65%
June 1%
July 0%
August 0%
September 3%
October 96%
November 91%
December 77%