Site hosted by Build your free website today!


The Guns of the Santíssimo Sacramento


This study is an application of technological analysis to marine archaeology by a military historian. It deals specifically with the cannon of the Portuguese galleon Santíssimo Sacramento, lost off the coast of Brazil in May 1668 and recovered by divers of the Brazilian Navy between 1976 and 1978. The study has three main purposes The first is to demonstrate the potential contribution of technical analysis of early modern ordnance to marine archaeology and the history of warfare at sea. The second is to extract the maximum benefit from archaeological investigation of the wreck. The third, and most important, is to place the early modern cannon founderhis understanding of his art, his control of his medium, and his relationship with the seaman and shipwright—in an appropriate historical and technological context.

The interplay between marine archaeology and other disciplines is hardly new. As a high-technology discipline habitually open to innovation, marine archaeology has been willing from the first to seek out and exploit specialist expertise across a wide range of fields. Examples of this practice are the use of exotic preservatives to stabilize timber recovered after millennia of immersion, the exploitation of radiocarbon techniques to date organic material, and the application of the skills of the numismatist and of experts in early pottery and glassware. The present study is an example of this openness of approach. It began with an invitation to me from the Historical Service of the Brazilian Navy, the Serviço de Documentação Geral da Marinha, to visit Brazil to examine and analyze the recovered ordnance .

Despite the recognized importance of guns and gunnery to early modern naval history and despite recognition of the importance of cannon recovered from wrecks of the 16th and 17th centuries, little effort has been made to apply the specialized study of ordnance to marine archaeology. Significantly, most of the analytical work done on cannon recovered from beneath the sea has been done by technologically oriented diver-archaeologists rather than by academic scholars. As a result, guns recovered from early modern wrecks have been effectively exploited in only a handful of cases, an example being Geoffrey and David Allen's work on the guns of a Portuguese India galleon lost off Africa in 1647 and also named Santíssimo Sacramento.#1 A general lack of understanding of early artillery, particularly naval artillery, is both a cause and consequence of this situation.

It is safe to say that the systematic study of early modern ordnance has hardly begun; similarly, what little is known has not been effectively applied to marine archaeology. The reasons for this are many and powerful. Marine archaeology draws on and feeds a general fascination with the development of ships, a fascination which increases in intensity as we go further back into history and into areas for which written records are scarce. Cannon, however, appear relatively late in the development of ships: the oldest wrecks having cannon associated with them date from the mid-16th century, by which time written records were relatively abundant. The Mary Rose, an English Galleon sunk in l545, is the earliest such wreck which comes readily to mind.#2

Much of the specialist knowledge applied to marine archaeology has been devoted to the dating and identification of wrecks. The wrecks of cannon-armed vessels however, are usually recent enough to be identified by archival research, an area into which marine archaeologists and treasure hunters alike have moved with alacrity. As a rule, the evidence associated with such wrecks is so extensive that there is no need to resort to an analysis of recovered cannon for dating or for identification. Where cannon are used, bronze cannon in good condition, with legible royal crests, dates, and founders marks, are normally involved. Since the information to be gained from cannon is apparently unnecessary, or easily obtained by other means, or both, there has been little pressure to expand our knowledge by in-depth analysis of the recovered ordnance.

The prevailing lack of curiosity about early naval ordnance is underlined by a tendency to assume that since artillery and the broadside sailing ship are comparatively recent developments, our knowledge of them is accurate and complete. Cannon went to sea in significant numbers only in the 16th century; by the middle of the 18th century, the manufacture and employment of naval artillery was rationalized, standardized, and organized along lines which seem familiar to us today. As a result, we tend to analyze ships and guns of the 16th and 17th centuries in terms of the technology, tactics, organization, and attitudes of the 18th and 19th. Even the cannon work against themselves: as early as the 1520s bronze cannon were being cast which were similar in overall appearance to cannon cast for the next three centuries.#3 Since there was little change in appearance, we assume that there was little change in substance. This implicit assumption is false.

An example of the tendency to read back the characteristics of later technology into 16th and 17th-century naval history occurs in the work of the prophet of seapower, Alfred Thayer Mahan. In his study of naval warfare under sail, Mahan persistently endowed the ships and fleets of the mid-1600s with the structural characteristics, firepower, seakeeping capabilities, and cruising endurance of those of the Napoleonic era. This led Mahan and those influenced by his ideas to focus their attention on the actions of the great battle fleets, holding that commerce destruction, or the guerre de course was at best an inherently wasteful strategy of doubtful value and that victory at sea could be reliably won only by a decisive major fleet engagement.#4 This belief, however, rested on the assumption that the fleets in question were capable of operating for extended periods and in all seasons at great distances from their home ports. That assumption is demonstrably incorrect for the 17th century, when the enormously powerful first and second-rate ships-of-the-line on which the main battle fleets depended were sadly lacking in both endurance and seaworthiness.#5 The campaigns of the battle fleets of the 17th century had a cyclic, seasonal character more reminiscent of 16th-century galley warfare than of the great naval campaigns of Nelson and his French and Spanish opponents.#6

Mahan's pervasive influence on the historiography of warfare at sea can be seen clearly, if indirectly, in the almost total lack of attention paid by naval historians to the extended 17th-century struggle over Brazil between Portugal and the Dutch, arguably the only truly decisive maritime war ever waged in the Western Hemisphere.#7 The Portuguese-Dutch conflict was not a war of great battle fleets; rather, it was a sustained war of economic attrition, fought well beyond the radii of action of the mighty ships-of-the-line that dominated the English Channel and contiguous waters. This has direct relevance to our analysis, for the intentions of the builders of the Santíssimo Sacramento were not in accord with Mahan's theories; she was built for a sustained war of convoy escort, raid, and counter-raid fought far from home, if not for the guerre de course itself.

The second example of reading back is even more to the point: generations of naval historians have assumed that long guns meant long range in the early modern period.#8 While this is true for modern artillery using nitrocellulose-based propellants, it is not true for black-powder cannon. The cannonball propelled by a charge of black powder reaches the maximum possible velocity at a point approximately 18 calibers (i.e., eighteen times the bore diameter) down the barrel. For reasons having to do with the internal ballistic properties of black powder—which are utterly different from those of nitrocellulose-based propellants in that the burning rate does not vary as a function of pressure or temperature—this maximum effective barrel length is, for all practical purposes, a constant.#9 Any additional length of barrel—and most 17th and 16th-century cannon barrels were considerably longer than 18 calibers—had no significant effect on muzzle velocity and thus no effect on range.

The tendency to attach importance to the supposed range advantage of long guns represents a third tendency to read later. conditions back into an earlier situation. Range, as the term is used today, had little practical relevance to early modern warfare at sea. Smoothbore cannon firing a spherical projectile were inherently inaccurate; the loosely fitting cannonball "balloted," or bounced, unpredictably down the barrel, acquiring "spin" in a random and uncontrollable fashion. As a result of this spin, the cannonball might slice or hook like a golf hall, or "float" erratically like a baseball pitcher's knuckleball. Attempts to hit anything beyond 500 yards or so were normally a waste of powder and shot, and long barrels—as we have just shown—had no more beneficial effect on accuracy than on range.

Not only was it impossible to hit a target with precision at long range, particularly from the rolling deck of a ship, but an inert iron cannonball also had very limited destructive powers at the end of its trajectory. There were occasional exceptions to this rule: a "long shot" might smash a spar or cut a crucial piece of rigging to disable an opponent and permit escape or the capture of a prize. But fascination with maximum ranges is by and large a modern preoccupation based implicitly on the characteristics of post-1850 rifled artillery firing explosive projectiles which are effective at maximum range.

There is, however, an area in which differences in barrel lengths were indeed significant in bronze cannon, one having nothing to do with muzzle velocity, range, or accuracy but, rather, with strength. In western European practice, cannon were cast breech down in a sunken pit, the molten bronze being poured into the mold through a so-called casting bell at the muzzle#10 (see fig. 1). The pressure developed by a column of molten bronze, like that developed by any liquid, is proportional to the height of the column. The casting bell served to lengthen the column and thus increase the pressure under which the molten bronze at the breech solidified (the casting bell was subsequently cut away). Even so, the greater casting pressure that could be achieved by increasing the length of the barrel further served to minimize the negative effects on the strength of the metal caused by the presence of impurities and by the porosity and sponginess inherent in bronze gunmetal over which the early cannon founder had little control. Tests conducted in the 1850s by Thomas Jefferson Rodman on cannon cast in the manner described above, though the guns were considerably shorter than those with which we are concerned, revealed that the metal at the breech was 5 percent denser than that at the muzzle. This modest increase in density increased the metal's ability to resist shearing stress, and hence its strength, by a factor of two.#11 Knowledge of this relationship enables us to evaluate the quality of early bronze ordnance, but whether the cannon founders of the 16th and 17th centuries were explicitly aware of it, we do not know. It is apparent, however, that they were implicitly aware of it, for they exploited it in a systematic and controlled manner, along with the other critical variables of cannon design.

Figure 1. Click to enlarge

We know, for instance, that the better founders cast their cannon within very close tolerances to a standard model which changed very slowly with time.#12 Good 16th and 17th-century ordnance of cast bronze set the standard for first-class artillery until well into the 19th century. We also know that there was explicit awareness of a positive relationship between barrel thickness and strength, though the nature of the relationship was not clearly understood—nor is it understood clearly today. We know, further, that cannon tended to get shorter and thinner as founders developed better control over the quality of their metal, and that such cannon were consistently cast by the better founders.#13

It may appear that we are using a circular argument, namely, that shorter, thinner cannon were better because they were shorter, and thinner. In fact, this is not so. Cannon cast by better founders, men clearly identifiable by their work if, not always by name, tended to show these characteristics and to be remarkably consistent in dimensions, weight, and the composition of their metal.#14 Such founders used their skill to minimize the amount of expensive bronze in their cannon; this resulted not only in cost savings but also in a lighter and handier piece for the same weight of ball. These advantages were obvious and were appreciated. Similarly, the penalties for casting cannon of insufficient strength were severe. First, the damage a burst cannon could do to a crowded gundeck—and a founder's reputation—was horrendous. Second, the cost of recasting the metal from a cannon which burst under proof (firing with an established overcharge) was sufficient to deter random experiments because the molds were constructed of fired clay and could be used only once. Minimum thicknesses and lengths were therefore as closely observed as maximums.

The length and thickness of a cannon are only two of a number of factors which must be considered in evaluating early ordnance; they tell us little about a cannon if we do not know its nationality and date of manufacture, for example. But taken in context, the relative length of the bore and the barrel-wall thickness, expressed as functions of the bore diameter, provide a useful and unequivocal, if nondimensional, indication of the cannon's quality and the founder's competence .

A final technical note is necessary before we turn to the analysis of the Sacramento and her guns. This involves the relationship between bronze cannon and cannon of cast iron, While there is no doubt that the development of a method for casting reasonably safe iron cannon, first mastered by the English and then by the Dutch, Germans, and Swedes, was an achievement of immense importance, this was not because of any improvement in the quality of the cannon. Cannon of cast iron were larger and heavier than cannon of bronze designed to fire a ball of the same weight. Worse, they were subject to internal corrosion and, partly as a result, were much less safe. When they burst they did not remain essentially intact as bronze guns usually did, parting like a torn sponge along a longitudinal line near the breech; rather, they blew apart in jagged fragments like a bomb. Iron cannon were used in considerable numbers, for they cost only a third as much as equivalent bronze guns, but they were not the weapon of choice.#15 Their use on a first-class warship of the 17th century suggests a serious shortage of ordnance.

* * *

Comparatively little documentary evidence has survived concerning the Sacramento. Chronicles and contemporary correspondence give the time and circumstances of her loss and little more.#16 When she went down on May 5, 1668, she was the capitânia, the flagship, of the escort provided by the Companhia Geral do Comerçio do Brasil for the annual Brazil convoy from Lisbon to Bahia and back. The annual convoy was of considerable economic importance to Portugal—it consisted of fifty ships in 1668—and the possibility of armed interference with it was considerable. There is thus no reason to doubt the word of the chronicler Sebastião de Rocha Pita when he states that Sacramento was "... one of the best ships in Portugal at that time."#17 The assignment as capitânia would have gone to just such a ship.

War between Holland and Portugal had gone on for decades along the Brazilian coast, in the East Indies, and on the sea-lanes between. Both sides had ample opportunity to test their mettle. We know from the frequent and well-documented results of Dutch encounters with the Spanish, English, and French that the Dutch were not lacking in courage, materiel, or skill in sailing and fighting broadside warships. But we also know that the Portuguese ultimately prevailed over the Dutch along the coasts of Brazil and, though our knowledge of the tactical details is scant, that victory owed much to the quality of Portuguese warships and the way in which they were handled. Beyond reasonable doubt, therefore, Sacramento was a first-class war-ship for her time and place.

Concerning the identity of her wreck, there is also no doubt. Documentary evidence relating the circumstances of Sacramento's loss coincides perfectly with the location of the wreck, at latitude 13° 12' 18" south, longitude 38° 30' 04" west, off the mouth of the Rio Vermelho, south of Salvador, Bahia.#18 Artifacts recovered correspond exactly to what we would expect from the wreck of a major Portuguese warship of the mid-1600s. More specifically, dates and founders' marks on the nineteen bronze cannon recovered from the wreck by Brazilian Navy divers under the supervision of archaeologist Ulysses Pernambucano de Mello e Neto (seven cannon were raised earlier by private divers under uncontrolled conditions) point clearly to a ship laid down around 1649 and launched no earlier than 1650, probably in 1653 (see table 1). All of this evidence meshes exactly with records of known Portuguese ship losses and points clearly to the Santíssimo Sacramento, lost in 1668.#19

Table 1. Click to enlarge

Significantly, the dates and founders' marks of the cannon recovered gave the first real confirmation of the ship's identity.#20 Decisive proof came with the recovery of artifacts bearing the personal monograms of the ship's captain, João Correira da Silva, and the captain general of Brazil, Francisco Correira da Silva, also lost in the wreck. Even with certain identification of the ship, we can say little about her without the aid of marine archaeology. We are therefore in debt to Professor Pernambucano de Mello e Neto, to the divers of the Brazilian Submarine Force, and to the officers and men of the submarine rescue vessel Gastão Moutinho from which exploration and recovery operations were conducted.

Contemporary sources are consistent in referring to Sacramento as a galeão, a galleon; one source states that she had sixty guns.#21 From this, we can deduce with certainty only that she was a purpose-built warship. While the term galeão can be associated with a particular type of warship in the 16th century (unlike the Spanish equivalent galeón, the term galeão was never applied to merchant vessels),#22 the middle of the 17th century was a time of flux and rapid change in naval architecture, and we cannot say with precision what the term meant when Sacramento was launched. Very little is known about the design, construction, and armament of Portuguese warships in this period. The sixty-gun figure may have been an artificial "rating" rather- than an actual count; it probably included numbers of swivel pieces, boat guns, and so on, which we today would not count as broadside cannon.

While a good deal is known about the design and armament of English, French, and Dutch warships of Sacramento's period, we cannot with confidence extrapolate our knowledge of them to Portuguese practice. This is in large part because much of the attention given to the development of warship design of the early to mid-1600s has focused on major ships-of-the-line, particularly massive, 100-gun behemoths such as Sovereign of the Seas, Prince, Soleil Royal, and Zeven Provinzien. These ships, like Santíssimo Sacramento, were considered the "best ships" of their time in their respective nations, but they were very different in concept and construction. They were also quite different from the English, Dutch, and French ships which contended with Sacramento and her sisters for the commerce and security of the Brazilian coast.

The Sacramento, a smaller and less powerful warship, might seem unimpressive in comparison with the huge battleships noted above, hut we must consider the strategic and technological context. The massive 100-gun ships of the mid-1600s mustered great combat power, but they were not true transoceanic warships. Unlike their equivalents of a century and a half later —Victory and the Ville de Paris are among the better-known examples—they rarely ventured far from home port and usually campaigned only briefly and during the relatively calm months of late spring and summer.#23 Enormously expensive to build and operate, they must be understood as the highly specialized craft they were; we must not read back into them the characteristics of the first-class ships-of-the-line of a later era.

In Sacramento's day, therefore, line-of-battle ships and genuine transoceanic warships were not identical. Ultimately, advances in naval architecture allowed the two functions to be performed by the same ship; the seventy-four-gun ships-of-the-line of the late 18th and early 19th centuries are the most notable and perhaps the most important examples of this. But Sacramento herself was a genuine transoceanic war-ship and not to be compared with ships-of-the-line of a later era.

It is therefore tempting to view Sacramento as a proto-frigate, a transoceanic cruiser which could either outfight or outrun any ship she was likely to meet as far away from home as the Brazilian coast. This is an intriguing hypothesis which is supported to a degree by the probable size of the galleon's gundeck and the weight of her ordnance.. All we can say with certainty, however, is that Sacramento's design was worked out according to the dictates of the peculiar tactical and strategic demands of the Companhia Geral do Comerçio do Brasil and within the bounds imposed by economic factors and the capabilities and limitations of the human resources available.#24

Since speculation concerning Sacramento's design and construction features is beyond the scope of this article, I will concentrate my efforts on her ordnance. Thirty-four cannon, twenty-six of bronze and eight of cast iron, were made available to me for examination; the bronze guns are listed along with their salient features in table 1. These guns are particularly important because they are representative of the best naval ordnance available to Portugal and because they have survived in relatively good condition. We have limited knowledge of an additional eight cannon of cast iron which were left on the bottom pending establishment of facilities for preserving them from the corrosion which attacks cast iron on exposure to the atmosphere after long saltwater immersion. We know the distribution on the bottom of most of the major items recovered, including thirty-five of the forty-two cannon accounted for. Though incomplete—for the location of the seven cannon recovered before archaeological controls were imposed is not known—the evidence provided by the plan giving this information is critical (fig. 2).

The distribution of wreckage in the plan suggests that the ship came to rest on the bottom right-side up on a relatively even keel. This disposition is plain from the arrangement of anchors and guns. The cannon were found in two ragged parallel lines. They were flanked by four of the five anchors at what we can safely assume was the forward end of the ship since a ship's main anchors were carried outboard in the bows. Deviations from this overall scheme are minor and reinforce the conclusion that the locations from which the cannon were recovered correspond closely to their locations on a horizontal plan of the ship before she went down.

The lines of cannon curve inward at the extreme stern just enough to suggest that the two cannon in the opposing lines closest to another were stern chasers, mounted side by side to fire rearward on either side of the rudder. The lines of cannon are least regular at the stern where the hull and superstructure would have been deeper, leaving a greater mass of rotting timber to disorder the rows of cannon in their slow trip to the bottom as the wreck decayed.

The length of the lines of cannon suggests a gundeck about 158 feet (49 meters) long. Such a gundeck length suggests an overall hull length of about 200 feet (61 meters) from stem to stern. An equivalent English or Dutch warship of. the Sacramento's vintage would most probably have been a third-rate ship mounting about forty cannon on her two main gundecks. The biggest of these cannon would have been no bigger than 32 pounders (i.e., designed to fire a 32-pound cannonball of cast iron) and probably no smaller than 24 pounders.

The guns of the lower gundeck, about half of the total, would have been larger than those of the upper gundeck by a considerable amount in terms of ball weight. If, for example, the cannon of the lower deck had been 24 pounders, we would expect 12 pounders, supplemented perhaps by the odd 18 pounder above.

Figure 2. Click to enlarge

This tallies closely with Sacramento's ordnance. Of the twenty-six bronze cannon recovered, two are very small pieces, 42 pounders which would have been mounted on the upper decks and are thus excluded from our analysis for now. The rest are split almost evenly between 20 pounders or larger- (twelve) and 12 pounders or smaller (ten). This and the close spacing of the guns in their two rows erase whatever doubt we may have that the ship's main battery was mounted on two decks, Two additional pieces, an archaic Portuguese 15 pounder and a long and unwieldy Dutch 15 pounder, might have been part of the lower gundeck because of their size and bulk; alternatively, they might have been assigned to the upper gundeck because of their relatively light shot.

The solution to this puzzle lies with the eight iron cannon still on the bottom. Of the eight iron guns recovered, four, judging by their gross external dimensions, fall into the 20 pounder or larger category and four fall into the 12 pounder or smaller category (this further confirms the two-gundeck hypothesis since the plan indicates that at least seven of the eight were found adjacent to one another in an area corresponding to the starboard quarter; i.e., the smaller guns of the upper gundeck fell through the rotting hull onto a like number of larger guns directly below them). The recollection of Capitão-da-Fragata Oscar Moreira da Silva, Gastão Moutinho's captain during the recovery operations, based on divers' reports, is that most if not all of the eight cannon remaining below were in the larger category.

This, combined with analysis of the guns recovered, suggests that the Sacramento's intended main gundeck battery was of bronze 26 pounders, but that a shortage of good bronze ordnance had led to the inclusion of numbers of cast-iron cannon and possibly also of bronze cannon as small as 20 pounders. Certainly, bronze guns were superior, and the Sacramento, a first-class warship on a mission vital to her nation's security, would have carried them had they been available. Data published by Carlo Cipolla in his seminal Guns, Sails and Empires suggest that there was a general shortage of good cannon in Portugal in the decades following the reassertion of independence from Spain in 1640.#25 This would seem to be a reasonable working hypothesis in light of Spanish dependence on Portuguese naval resources during the years of Habsburg dominance. The heterogeneous nature of the Sacramento's gundeck batteries constitutes strong evidence in support of this hypothesis.

Figure 3. Click to enlarge

The Sacramento's builders would undoubtedly have preferred a main battery of good Portuguese bronze ordnance; of that there can be no real doubt. This is confirmed by examination of the Portuguese cannon which we can definitely assign to the lower gundeck on the basis of size. There are eight of these: a 28 pounder by the founder A. G. F., probably Antonio Gômes Feio,#26 the largest cannon recovered (fig. 3); a 24 pounder by an unknown founder (though unsigned, the gun is plainly Portuguese in design and proportions and carries the royal crest on the muzzle); and six 26 pounders by the founder Rui Correa Lucas Matias Escartim, five of them cast in Lisbon in 1649 and the other in 1653, the latest date on any cannon recovered (fig. 4). These last six guns in particular are remarkable. There is a consistency of line and proportion which shows clearly that Portuguese founders followed an established model; comparison with the Sacramento's captured Dutch cannon (of which there are two in addition to the two 42 pounders and the 15 pounder mentioned earlier—see fig. 5) suggests that it was also a superior model. Though the three larger of the Dutch guns all fired a smaller ball—20, 20, and 15 pounds, respectively—they are heavier or longer, or both, than the Portuguese 26 pounders just mentioned.

Figure 4. Click to enlarge

Figure 5. Click to enlarge

The weight and size of a cannon as a function of projectile weight are, as I have suggested, unequivocal indications of quality. The way in which they are determined is therefore clearly central to this analysis. The external dimensions of Sacramento's cannon were measured with a steel tape, and their lengths and circumferences were recorded by me. While there are certain inherent inaccuracies in this method—parallax and the sag in a stretched tape, to name two—I considered these to be sufficiently small as to be practically irrelevant. It was further felt that consistency in technique would make the results valid for comparative purposes. This proved to be true only to a degree. The results are undoubtedly sufficiently accurate for comparative purposes, but the dimensional precision of at least some of the cannon came as a surprise.

Measurement of the six pieces by Lucas Matias Escartim revealed remarkable dimensional uniformity. In spite of the fact that each cannon was cast in an individual mold which was destroyed to remove the finished gun, and in spite of the fact that one of the sis was cast four years after the others, in essential dimensions they are extremely close copies of the same model. The maximum measured variation in length between any two of the six was some two-thirds of an inch (1.5 centimeters) from the muzzle to the base of the breech reinforce, a distance of 9½ feet (2.90 meters). For bronze castings of over 3,500 ponds (1,600 kilograms), such uniformity is remarkable, even by modern standards. Barrel-wall thickness, though impossible to compute precisely until the bores are cleaned of marine growth and measured exactly, shows evidence of similar precision.

Based on measurement of bore diameters at the muzzle, a process requiring estimation since wear has rounded the inside corners, there is a maximum variation in the thickness of the barrel wall immediately behind the touchhole—probably the most critical single dimension—of only 3 percent. If. we assume that two of the cannon with particularly worn muzzles whose bores were measured as being larger by about .04 inches (2 millimeters) were, in fact, 26 pounders with the same bore diameters as the others, the difference drops to 2 percent.

Less critical dimensions were apparently less closely controlled. The trunnions of two of the six Lucas Matias Escartim pieces are noticeably skewed in the horizontal plane, an inconsistency also present on one of the six Portuguese 11 pounders (fig. 6).

Lucas Matias Escartim's success in achieving dimensional conformity in his cannon was echoed in his success at maintaining only a small variation in their weight. The heaviest of his six cannon weighs only 1.42 percent more than the mean, the lightest only 2.93 percent less, despite the documented and considerable difficulty in standardizing barrel weights, a problem which was never soled as long as bronze cannon were cast.#27

At this point a more detailed explanation is in order concerning our use of weights because weight, as we have seen, is a crucial factor in the evaluation and comparison of cannon. The cannon have not been weighed following their recovery, scales capable of accurately weighing objects of 3,500 pounds (1,600 kilograms) not being commonplace. Our only positive evidence of the weights of the cannon is therefore the markings stamped on their breeches.

We have treated the incised weight markings in the form a36I16a as indicative of the weight of the barrel in quintaes (hundredweights), arrobas (fourths of a hundredweight), and arratels (Portuguese pounds).

Figure 6. Click to enlarge

Figure 7. Click to enlarge

Figure 8. Click to enlarge

This assumption was validated by Sacramento's six English cannon which bear, in addition to Portuguese weight markings, English markings in the form 2630A (figs. 7 and 8). Markings in this form plainly represent barrel weight in pounds avoirdupois which had a constant value of 7,000 grains, or 453.6 modern grams, for the entire period in contention.#28 We can therefore use the double markings on the six English cannon to test the validity of our assumption concerning the nature of the Portuguese weight markings. We can also use a comparison of the two sets of markings to establish the Portuguese unit of weight and to gain some idea of the prevailing standards of precision and accuracy in Sacramento's day.

A least-squares regression analysis shows that the two sets of markings are parallel expressions of the same quantities in different, but consistent, units of, measure with a correlation of .9989 (see Appendix A), confirming beyond any reasonable doubt that the three-part markings on the breeches of Sacramento's cannon are indeed weight markings. Our assumption was yet further confirmed by computations of the cannon's expected weights based on their estimated volumes and the density of their bronze.#29 Similarly, the small inconsistencies between the two sets of markings when mathematically converted to a common unit make it clear that the Portuguese did not simply accept the English weights and multiply them by a conversion factor without actually weighing the cannon.

The agreement in the findings suggests that unexpectedly high standards of accuracy and precision were observed by English founders and Portuguese arsenal workers. When we consider that the correlation was adversely and cumulatively affected by the inaccuracies in each of two weighings, this exactness is particularly impressive. It is also significant because it implies that the weights marked on the guns may well have been used in English and Portuguese practice to assist in balancing ship. Support for this theory is offered by the fact that Dutch cannon were not marked with their weights at the foundry, yet the two largest of Sacramento's Dutch guns have what appear to be Portuguese weight markings roughly scratched onto them (table 1; the third element, denoting arratels, is omitted on the Assuerus Koster 20 pounder). If the guns captured by the Portuguese were weighed on the spot with field scales, however incapable of fine precision, the practice implies a high standard of competence and care in preparing and arming ships for war.

The double weight markings on the English guns raise a final question. If we discard one pair of markings which is slightly "out"—the correlation now goes up to .9999—we derive a value of 465.8 grams for the arratel, a value which corresponds to no known value of which the author is aware, 459 grams usually being given.#30 This finding argues that there may have been a special naval arratel or perhaps a secret ordnance arratel.

Before proceeding, I want to make a point concerning consistency—or the lack thereof—in units of weight. For convenience, I have given the weights of cannon in the original Portuguese units. My purpose is comparative, and, so long as I am consistent, Portuguese arratels will serve as well as any other unit. But I have given the "ratings" of the cannon, that is, the weight of ball they were intended to fire, in pounds avoirdupois in order to facilitate comparison with the armament of contemporary ships of other countries. Virtually all published references to early modern naval armament categorize cannon in terms of ball weight in pounds avoirdupois, and the difference is in fact small. This raises the possibility, addressed in Appendix B, that the Sacramento's 26 pounders discussed above were intended to fire a ball of 25 rather than of 26 arratels. This need not deter us here except to note that the standards of precision which permit speculation on this point lend confirmation to my conclusions concerning the high quality of the best Portuguese ordnance.

To form a definitive opinion of the quality of Lucas Matias Escartim's pieces, we will have to determine the degree to which their bores approximated perfect cylinders and the precision with which they were centered. As yet, it is not possible to do so. Nevertheless, the relative shortness and lightness of these six pieces—their bores are just a shade over the 18 caliber length which represents the point of diminishing internal ballistic returns—leave little doubt about their excellence.

Their quality is echoed in that of the Portuguese 11 pounders of the upper gundeck, though with intriguing differences. There are six 11 pounders of Portuguese origins among Sacramento's guns, but of the six only two are of a vintage and apparent quality comparable to that of the six Lucas Matias Escartim 26 pounders. Though they bear no founder's mark, these two 11 pounders are clearly closely related to the larger guns. The arrangement, spacing, and contours of the reinforces and cascabels are virtually identical to those of the Lucas Matias pieces; the royal crest and monogram of Dom João IIII are rendered identically, and the design of the dolphins on the barrel tops and breeches is virtually identical as well.

Given the necessary difference in proportions between an 11 pounder and a 26 pounder, it is apparent that the two 11 pounders in question and the six Lucas Matias pieces were made within the same manufacturing tradition and probably in the same foundry. The smaller pieces probably bear no founder's mark because the casting of larger cannon is considered more important and prestigious and was supervised by the master founder himself; while the casting of smaller pieces was entrusted to understudies or apprentices. This hypothesis is reinforced by the fact that the only one of the Sacramento's Portuguese 11 pounders to bear a founder's mark, that of the founder A. G. F., is the lightest of the six by some 150 pounds (68 kilograms); it was probably cast at least a decade earlier, when smaller cannon still presumably commanded a master founder's attention.

A direct comparison between the foundry practice of A. G. F. and Lucas Matias Escartim, based on the A. G. F, 28 pounder, suggests that the two founders were of comparable ability, The weight markings on the breeches indicate that the larger A. G. F, piece contains just over 140 pounds of bronze for each pound of cannonball thrown, while the six Lucas Matias pieces contain from about 140 pounds to just under 142 pounds. Like the Lucas Matias 26 pounders, the two similar 11 pounders show a strong dimensional uniformity; they vary in length by only ½ inch (1.3 centimeters) in 9½ feet (2.9 meters) and in weight by less than 60 pounds in over 2,500 (1,134 kilograms). One of the two, like two of the six larger pieces, has its trunnions skewed in the horizontal plane by about 1½ degrees.

The remaining three of Sacramento's Portuguese 11 pounders are a mixed bag. They all appear to be older than the other three. They do not, however, differ dramatically from them in proportions or weight. The Portuguese had apparently found cannon of this size and ball weight to be useful well before the mid-1600s and had standardized on them to the degree possible. If our galleon's gundecks are an accurate indication—and it should not be forgotten that Sacramento's cannon are fixed archaeologically in time and context—the English may have standardized along similar lines, for three of Sacramento's total of nine bronze 11 pounders are English.

Significantly, if we assume an equivalent level of technology, smaller cannon tend to be heavier in terms of projectile weight than larger ones. The difference in relative weight was not trivial. The six 11 pounders range from some 221 pounds of barrel per pound of ball (the A. G. F. piece) to over 242 pounds, all containing almost 100 pounds of bronze per pound of ball more than the 26 pounders. The most likely reason for this relative inefficiency is suggested by the relatively greater lengths of the smaller pieces. Though designed to throw a ball weighing less than half as much, the 11 pounders are only marginally shorter in absolute terms than Sacramento's 24, 26, and 28 pounders. This was probably due to the need we have already discussed to cast a column of molten bronze of a certain minimum height in order to produce metal of the density and strength needed for a cannon's breech.

Why, therefore, were these relatively inefficient smaller guns cast at all? In land use, the rationale for greater numbers of smaller guns, as opposed to a few larger ones, however ballistically efficient, is clear. The inflexible restrictions of horse traction placed obvious limits on the mobility of large cannon, and several small projectiles were tactically more effective than a single large one of the same weight when engaging dispersed human and animal targets. But at sea, where destruction of a ship was the main objective, the advantages of larger guns in terms of ballistic efficiency and relative cheapness would seem to have been compelling. Whatever advantage smaller cannon might have had in rapidity of fire would have been more than offset in their relative lack of destructive impact.

On reflection, it is apparent that the issue was more complex, revolving around such sophisticated questions as the strength and weight of bulwarks and gundecks, center of gravity considerations, and moments of inertia—though these were not explicitly understood for many decades. The naval architects who designed and built Santíssimo Sacramento undoubtedly had clear ideas concerning the preferred size, composition, and arrangement of her main and upper gundeck batteries. It is likely, therefore, that 26 pounders below and 11 pounders above represented a ballistic and structural ideal, the optimum combination of useable firepower which could be built into a truly transoceanic warship in Lisbon—or anywhere else—in the 1640s and 1650s. Does Sacramento's varied assortment of 15 and 20 pounders (of which there were no less than six, all of them but a single long, heavy, and probably old Portuguese 15 pounder either Dutch or English) therefore represent convergence toward the ideal or the acceptance of limited supply? The absence of first-class Portuguese cannon in the 15-20-pound category strongly suggests the latter., but we simply do not know.

In this context, Sacramento's English cannon take on added importance. Four of the six were cast by founders known to have worked in England at times corresponding to the dates cast into their breeches alongside the founders' names, the dates ranging from 1590 to 1597.#31 The other two, though neither signed nor dated, are English in form and markings. They appear to be much older than the others, corresponding in size and shape to demi-culverins from the wreck of the Mary Rose the Portuguese royal crest is cast into their muzzles, apparently dating them before the incorporation of Portugal into the Habsburg Empire in 1580-81, but their overall appearance is older still.#32 Careful comparison of their lines with those of mid-16th-century English cannon of known vintage in museum collections, particularly that of the Museum of the Royal Artillery Institution at Woolwich, leads to the inescapable conclusion that they were over a century old when Sacramento was launched.

The age of these English cannon is undoubtedly indicative of a general shortage of good ordnance. The efficient proportions and apparent high quality of the new Portuguese 26 pounders and 11 pounders make it clear, prima facie, that more of them were not used only because no more of them were available. But we should not make too much of this, for the most persuasive evidence of shortage which the English guns provide is not their age but the fict that three of the six deviate from the presumed standard of 26 pounders above and 11 pounders below,

If their condition after three centuries of saltwater immersion is any indication of quality, the Sacramento's English cannon were every bit as well made as the rest of the ship's ordnance and a good deal better than some of it. Despite their advanced age when the ship went down, they are in better condition today than all but the Lucas Matias Escartim pieces and two of the Dutch guns, More significantly, the four dated English cannon are comparatively light in terms of projectile weight.

The two 20 pounders by John and Richard Phillips (or Phellipps) (see table I) compare favorably with the only other 20 pounders aboard; these are two Dutch cannon, one cast by Assuerus Koster in 1654 and the other by Conrad Wagwaert in 1649. if we assume that the rough inscriptions on the Dutch guns referred to earlier represent their weights, we get a value of approximately 190 pounds of gun per pound of ball for the Dutch cannon and 180 pounds per pound for the English. When we consider that thirty-eight years of development separated the newest of the English cannon from the oldest of the Dutch and that the whole thrust of cannon development was toward shorter, lighter guns (a point thoroughly documented for English cannon by Michael Lewis),#33 the quality of the English guns is apparent. This point, and the impression of quality, is further underlined by comparing the two English 20 pounders made by the Phillips brothers with each other (see fig. 7). The newer of the two, though clearly made to the same basic model, is significantly thinner toward the muzzle and 30 pounds lighter. Though a sample of two and a difference of 30 pounds is too thin a statistical basis for sweeping generalizations, it is hard to escape the suspicion that we can see, in these two guns, a process of incremental modification under extremely close quality control, directed toward the ideal of lightness and shortness. The virtually identical proportions of the two 1597 11 pounders by George Elkine strengthen the suspicion.

The presence in Sacramento's battery of five Dutch cannon, like that of six English cannon, further implies a shortage of good ordnance. The reasons, however, are different. Unlike the English guns, the Dutch cannon are not particularly old. The oldest was cast in 1622 (the undated 42 pounder by Assuerus Koster was Probably cast in the 1630s or 1640s). Yet they are utterly unstandardized in bore diameter and ball weight (see fig. 5). Since they were probably acquired through capture, this is understandable; but the need to use a polyglot mixture of captured cannon on a first-class warship speaks for itself unless the ordnance in question was of a clearly superior quality, and there is nothing to suggest that it was.

The possibility that the Dutch and English cannon were ballast or booty picked up on the outbound leg from Lisbon, carried in addition to the Sacramento's gundeck batteries, seems remote in light of the disposition of the recovered cannon on the bottom. Had they been carried as cargo or ballast, they would have been stored in the hold and thus found clustered together near the center of the ship. The possibility that they were captured on the outbound leg and used to fill empty gunports, though it does seem very remote, cannot be totally excluded. And if this were the case, it would even further substantiate the hypothesis that good ordnance was in short supply.

If we leave the two small deck pieces out of the discussion, for we have nothing against which to compare them, the two Dutch 20 pounders match only the two pieces by the Phillips brothers. The Dutch 15 pounder finds a counterpart in the Portugese piece by the founder P. D. B. (fig. 9). The proportions of these two cannon are very similar, but this apparently favorable comparison must be qualified because the P. D. B. cannon, though undated, is probably the oldest of Sacramento's Portuguese cannon. It is longer and thicker in its proportions than the other Portugese cannon and differs markedly from them in both the design of the breech cap and cascabel and in the color and condition of its metal, which is badly pitted and has a distinct greenish hue. The simple and relatively small Portuguese royal crest on the breech of this piece is nearly identical in design to the crests on the two oldest English cannon. It is possible to infer from this evidence that the P. D. B. gun was cast before 1580. The implication is that for both guns the Dutch founders followed a model which had already been abandoned as obsolete in England and Portuguese. The considerable variation in color among the Dutch pieces—the Conrad Wagwaert 20 pounder has a blackish, almost ebony-like sheen, and the Henricus Meus 15 pounder has oxidized to a light pastel green—suggests that their Dutch founders had not yet established the same degree of control over the composition of their alloy that English and Portuguese founders had achieved.

It is, however clear that the Dutch cannon differ markedly in design and construction from their Portuguese equivalents and were comparatively longer and heavier. There is also reason to believe that Dutch methods, whatever their technical merits, produced a gun which was more expensive than its Portuguese equivalent.

This is perhaps unfair to the Dutch founders. With the exception of the Conrad Wagwaert 20 pounder cast in 1649, the Sacramento's Dutch cannon are older than all but the oldest one or two of the dated Portuguese guns and the Wagwaert piece is only marginally longer and thicker than we could expect a contemporary Portuguese 20 pounder to be. It is worth noting in this context that while the Sacramento's 1634 20 pounder by the Dutch founder Assuerus Koster is relatively long, thick, and heavy by Portuguese standards, a 1649 28 pounder by the same founder in the Naval Shipyard in Rio de Janeiro is close in proportions to the best of Sacramento's 26 pounders. This suggests that Dutch founders had caught up by mid-century. The fact remains, however, that Sacramento's Dutch guns are longer and thicker than her Portuguese and English equivalents despite the fact that the English cannon are from a quarter to a half-century older.

Figure 9. Click to enlarge

In one area in particular we have clear evidence of Dutch economic inefficiency. Even if we assume a rough parity in ball weight to barrel weight ratios, Dutch cannon must have been significantly more expensive than their Portuguese and English equivalents in terms of hours of skilled manpower expended per pound of ball thrown. The Dutch cannon are, to a gun, encrusted with elaborate raised floral ornamentation, inscriptions, and nautical motif. The presence of elaborate decorated guns as functional booty on the gundecks of an enemy ship suggests convincingly that such ornamentation was not confined to a handful of select presentation pieces. Even ordinary operational Dutch cannon, as attested to by the cannon of the Sacramento in particular and by recovered guns in general, were showcases for the low-relief sculptor's art. The beauty of the results speaks for itself, but ornamentation added weight, absorbed labor, and did nothing whatever for ballistic efficiency. Even if reusable mold parts were used for the decorated portions of the barrels (which seems unlikely, since most 17th-century Dutch cannon of which I have knowledge are one mass of extended decoration over large portions of their surface), the work involved in attaching, joining, and sealing the reusable parts would have been above and beyond the work of making the basic tube of the mold.

Why did the Dutch expend such effort on ornamentation with such an apparent disregard for weight and cost One explanation may be that labor was cheaper in Amsterdam and The Hague than in London or Lisbon. This hypothesis is, however, questionable on both technological and economic grounds. The transition from stone to cast-iron a cannonballs demonstrates that when Sacramento's Dutch cannon were made, the cost of labor in northern Europe was, in fact, high and had a significant effect on gun-founding practice. Northern European founders were among the first to abandon cannon designed to fire cannonballs of cut stone. The main impetus for the abandonment of stone cannonballs and cannon designed to fire them was economic, the rising cost of labor. Stone cannonballs were more effective tactically than those of cast iron, particularly at sea. Cannon designed to fire them used a third less bronze for the same projectile weight, a significant economic advantage since bronze was expensive. But the labor of a skilled stonecutter—and immense skill was required to cut perfect spheres of stone to precise dimensions—was expensive too, and wage rates rose precipitously beginning early in the 16th century and continuing through the 17th.#34 Significantly, the stone-throwing cannon fell from favor earliest in areas where the wage-price spiral danced earliest. Northwestern Europe was the first such region. Where economic conditions permitted, the technical and tactical advantages of stone-throwing cannon continued to be exploited. The Portuguese cast cannon designed to shoot stone cannonballs in India long after they had abandoned them in the West. The most recent Portuguese stone thrower of European manufacture in the collection of the Museu Militar in Lisbon was cast in 1578,#35 while such cannon were cast in India through the middle of the 17th century. Stone-throwing cannon continued to be cast in the Ottoman Empire, where the wage/price spiral was much slower than in northwest Europe, into the 18th century.#36 In sum, there is compelling evidence that labor, both in general and as it affected cannon founding, was relatively expensive in northwestern Europe in the middle 1600s, yet Dutch cannon founders persisted in the obviously labor-intensive practice of elaborately decorating their product. Why?

Are we dealing with an entirely distinct technological ethos, involving an independent and separate tradition of cannon founding? Were Dutch founders, to overstate the point, romantic exponents of an obsolete tradition, while their English and Portuguese brothers were hardheaded realists, held firmly in touch with technological efficiency by fiscal reality.

While these questions must remain speculative, it is certain that Dutch constructional techniques represented in certain of Sacramento's cannon differ sharply from contemporary English and Portuguese practice in at ]east one important particular. Following a tradition which can be traced back to Biringuccio's Pirotechnia of the 1530s, Sacramento's English and Portuguese cannon were cast with a device to center the core of the mold. The device, called a cruzeta, was of wrought iron. It consisted of a thin ring with an inside diameter equal to that of the bore, held in the center of the mold by evenly spaced rods, usually four, projecting outward from it in the same plane#37 (see fig. 1). When the cannon was cast, the cruzeta remained imbedded in the cannon metal. For this study, investigation of the cruzeta was limited to efforts to locate the external tips of cruzeta rods. Where the rods had corroded away, holes—typically round and ¾ inches (1.9 centimeters) in diameter—could be seen. Where they were not visible, a magnetic detector was used.

Remarkably few of the cruzetas have, in fact, corroded away; those which have not are hard to see—in some cases, impossible. Their magnetic attraction is, moreover, slight. This presented few problems with the more modern Portuguese 26 pounders and 11 pounders. The location of the cruzeta tips along the fore-and-aft axis of the barrel was consistent, permitting the area of magnetic search to be narrowed down. On all of these cannon the angular placement of the four cruzeta tips in a transverse plane corresponded closely to 1:30, 4:30, 7:30, and 10:30 o'clock, when the cannon were viewed from the rear and the breech was seen as a clock face. Variations in the radial placement of the tips from this ideal scheme were typically less than thirty minutes on the imaginary clock face and as much as an hour in only a few cases. This incidence of variation suggests that cruzetas were individually fitted after the molds had been set up, implying the expenditure of considerable effort to insure that the cores were precisely centered. Indeed, the founders knew well that barrel-wall thickness at the breech was the most critical dimension in a cannon.#38

With the older Portuguese cannon, cruzeta placement was less standardized. In several places it appears that the cruzetas had only three legs, though it is probable that at least some of the missing "fourth legs" could not be found because they were at the bottoms of cannon breeches which could not be lifted to check. The cruzeta tips on the English cannon were particularly difficult to locate; as a result, it is possible that the Phillips brothers used a design with only two horizontally opposed supporting rods.

Nevertheless, it is clear that the founders who produced Sacramento's Portuguese cannon and at least her four newest English guns worked within the same tradition of cruzeta design and placement. The Dutch cannon—two of them, anyway—are another story. The small 42 pounders appear to have cruzetas much like those of Sacramento's English and Portuguese pieces. This is possibly true also of the 1649 20 pounder by Conrad Wagwaert; although only magnetic indications were found and in only one spot, that spot was consistent with Anglo-Portuguese practice. The other two Dutch guns, however, were cast with some sort of internal iron structure within the cannon metal about halfway down the barrel. This is clearly apparent on the 1622 piece by Henricus Meus (fig. 10) which has two square holes measuring about 1 x ¾ inches (2.5 x 2.0 centimeters) on top of the barrel, one forward of the touchhole and the other between the dolphins. The 1634 20 pounder by Assuerus Koster has two conventionally placed cruzeta tips at 10:30 and 1:30, but there is also magnetic evidence of a mass of ferrous material beneath the surface between the dolphins. This last indication is particularly puzzling: the cannon in question has been well cleaned and polished (it is kept under cover at the Second Naval District Headquarters in Salvador), so there can be no question about the quality and appearance of the surface of the gun, yet there is no visual or tactile evidence whatever of anything but a uniform bronze surface in the area of strong magnetic indications.

Was the 1634 Assuerus Koster piece cast with a free-floating structure of iron within the barrel? Or was it cast with an iron structure protruding into the mold which was cut away after casting and filed down into a depression which was subsequently filled with molten bronze? The first hypothesis seems unlikely on structural grounds. The second seems improbable in light of the labor and extremely hard and abrasive cutting elements required. All we can say with certainty is that some Dutch founders used a second cruzeta-like structure embedded in the cannon metal approximately halfway down the barrel and that in some cases it either reached or approached the surface of the cannon at the 12:00 o'clock position between the cannon's dolphins.

Clearly the Dutch founders represented on Sacramento's gundeck embedded more iron beneath the surface of their cannon than their English and Portuguese contemporaries, It is possible that the second cruzeta allowed them to dispense with external supports for the mold core. It may also have allowed them to dispense with the use of a casting hell, yielding savings in bronze and fuel required for casting. This hypothesis squares with the relatively greater length of the Dutch pieces, for a longer barrel served the same purpose as the casting bell, ensuring denser, stronger metal at the breech. Yet such a practice would ultimately have been inefficient in terms of bronze consumption, for the height of the casting bell could be varied independently without affecting barrel proportions, and the bronze in the bell could be remelted and reused after it was cut away.

Figure 10. Click to enlarge

The only plausible economic advantage in such a casting method would have been in fuel consumption, a smaller amount of bronze having to be melted for each cannon. Although it is difficult to imagine this advantage being important except in a very small-scale operation, the cost of the fuel involved, charcoal, may be relevant. There was a relatively unified world market in bronze, and in its constituent metals, copper and tin; all were high-value, low-bulk commodities which could be economically traded over long distances. But the Netherlands was presumably no better off in charcoal resources than England, and we know that England effectively exhausted her industrial charcoal resources shortly after 1600 (one of the most important effects of this development was the elimination of England as a factor in the international export trade in artillery).#39 Though the Dutch could presumably have imported fuel from Germany and elsewhere, the impact of freight costs on the cost of charcoal, a high-bulk, relatively low-value commodity, would have been considerable. Still, the Dutch were specialists in high-bulk, low-value maritime trade and were able to sustain a generally favorable trade balance. The high cost of fuel was probably, therefore, of no great importance to Dutch founders and the facts most relevant to our analysis remain the relatively inefficient proportions and the apparently inefficient design of Dutch artillery.

The two oldest of Sacramento's English guns offer evidence to suggest that the Dutch were latecomers to cannon founding, using technically effective but economically inefficient technology as they caught up. Both of these cannon show, like the Dutch guns, evidence of large amounts of ferromagnetic material embedded in their bronze, in this case in the breech caps, the trunnions, and the lifting rings on top of their barrels, all areas in which high strength was desirable. Though there is no written record of composite construction, it is possible that these cannon were cast with a wrought-iron reinforcing structure interlacing the bronze and that the Dutch guns with indications of an internal iron structure were similarly made. The long persistence of nonfunctional "reinforcing rings" on bronze cannon barrels suggests a continuity of design tradition extending back to the wrought-iron bombards of the 15th and 14th centuries with their thoroughly functional shrunk-on hoops. The composite iron and bronze construction of Sacramento's oldest English cannon may well represent a missing link in our tradition of design and manufacture.

We know that there were traditions of cannon funding other than the cine we have described here and traced back to Biringuccio. Perhaps the best documented of these competing traditions was that used by the Ottomans in casting large stone-throwing cannon. These guns were usually cast with their breeches uppermost in the casting pit, an arrangement which permitted relatively thin barrel walls forward of the powder chamber and resulted in significant economies in the amount of bronze required for the weight of ball thrown.#40 This method remained viable for as long as labor was cheap and bronze expensive and continued to be employed in the Ottoman Empire well into the 1700s, Ottoman cruzeta placement on iron-throwing cannon also seems to have varied, in some cases, from the practice we have described here, no less than five cruzeta legs being observed on the surface of the breech of one example.#41 At least some Ottoman stone throwers were cast breech down, or so the existence of cruzetas in their breeches would seem to indicate.

Other traditions of bronze cannon design and founding may well remain undiscovered, perhaps including a common hut as yet undiscovered ancestral tradition which was superseded for economic reasons, Further analysis of early bronze cannon throughout the world would be a first step toward developing hypotheses concerning the origins of cannon and gunpowder, an area which, despite its historical importance, has been the subject of far more uninformed speculation than serious investigation.

* * *

Thus the analysis ends where it began, with the importance of early ordnance in general and early modern naval cannon in particular. Through analysis of the guns of Santíssimo Sacramento's gundecks I have reached a number of conclusions that both validate some old hypotheses and suggest new approaches to the analysis of early guns and gunnery.

We now know that at least the very best early modern bronze ordnance could remain in active service at sea much longer than we might have supposed, perhaps for as much as a century and a quarter. We suspected this previously but lacked the unequivocal supporting evidence provided by Sacramento's two archaic English cannon. We cannot say how exceptional these cannon were, but their, presence aboard an operational warship well over a century after their manufacture, positively confirmed by the archaeological context, tells us a great deal more than we would have suspected in their absence.

It suggests, for instance, that the very finest bronze cannon of the 16th century were barely inferior, if at all, to the best cannon of the 17th and were certainly better than the run-of-the-mill average. This contradicts the popularly held and perhaps dominant view of technology as advancing by qualitative improvements in capabilities and characteristics. It implies that evidence of advances in cannon technology should be sought in relative reductions in size or weight rather than in literary evidence. It suggests that reductions in cost may have justified sacrifices in operational performance on a regular basis.

The long survival in operational service of cannon with archaic features suggests, in short, that earlier foundry practice may have produced technically superior ordnance by labor-intensive methods which could not be retained in the face of the wage and price spiral of the late 16th and 17th centuries. Long survival in demanding service at sea is eloquent testimony of the cannon's technical quality; the disappearance of the constructional methods used in their manufacture was therefore almost certainly unrelated to technical excellence or tactical effectiveness. The most reasonable hypothesis is cost and the most likely causal mechanism the rising price of labor, a factor which was offset, in part if not wholly, by economies of scale as capital investment increased and cannon founding made the transition from a craft to an industry.

The degree of control which the best of the Portuguese and English founders represented on Sacramento's gundecks exercised over the physical characteristics of their product suggests that historians of technology and science have badly underestimated the early modern cannon founder.#42 The evident care and precision with which the English and Portuguese weighed their naval ordnance suggest that the early modern sailor, shipwright, and gunner have been similarly underestimated as well. Their work has not based—at least so far as we know—on elegant theories of internal ballistics, metallurgy, or the relationship between stress and strain in thick-walled tubing; nevertheless, their application of incremental development based on trial and error supported by close quality control was highly successful. And we still have no satisfactory theoretical explanation for their success.

It is, however, apparent that the development of cannon founding did not proceed evenly throughout Europe or the rest of the world. Of the known traditions of foundry, the English seems to have advanced furthest during the 16th century, perhaps in concert with the Portuguese. The Dutch, in contrast, seem to have lagged behind the Portuguese during at least the first half of the 17th century, preserving an independent foundry practice tradition anti producing cannon which were longer and bulkier than their Portuguese equivalents and probably more costly to produce in terms of man-hours of labor as well.

The belief that Portugal was desperately short of ordnance following her reassertion of independence in 1640 is strongly confirmed by the cannon of Sacramento's gundeck. Despite evidence that Portugal's founders and shipwrights were well aware of the value of artillery standardization and were technically capable of achieving it, the Sacramento carried an amazing diversity of armament. Nearly half of her gundeck ordnance was of cast iron and much of her bronze ordnance clearly represents what was available rather than what was preferred. The demonstration of Portuguese technical competence offered by Sacramento's cannon makes the evidence of shortage even more persuasive.

Virtually everything we have learned from the Sacramento's gundeck is heavily colored by the archaeological context. This fact accounts for our most important conclusion. Had our cannon been preserved independently and studied individually, out of the context offered by marine archaeology, our findings would have been quite different and far more tentative and limited. It is no exaggeration to assert that the application of marine archaeology broadens the study of early modern ordnance from a technical exercise of limited value to a potentially powerful source of evidence for the economic and social historian as well as for the student of naval architecture and weaponry.

Appendix A. Click to enlarge

Counting All Six Cannon

Average uniformity = .97867 pounds per arratel, yielding a 463.5-gram arratel. Correlation coefficient = .99893.

Discarding Number 13

Average uniformity = .97385 pounds per arratel, yielding a 465.8-gram arratel. Correlation coefficient = .99992.

The correlation coefficient is a measure of the degree of consistency in the values of the weights of the individual cannon in one unit of measure expressed in terms of the other. Put another way, if the correlation coefficient were 1.0000, dividing the English weight of each of the six cannon by its Portuguese weight would yield exactly the same value in each case and the computed grams per arratel figures would be identical.

Mathematically, the correlation coefficient, r is expressed by the equation#43

where m is the slope of the curve of the weights of the cannon in pounds, x, plotted against their weights in arratels, y, on a two-dimensional scatter diagram, and where s represents the square root of the variance of the weights in pounds, sx, or arratels, sy, according to the expression

sy2 = variance of the weights in arratels

Appendix B

Units of Measure, "Windage," and the Categorization of Sacramento's cannon.

Uncertainty concerning the units of measure used in weighing the Sacramento's cannon and a lack of unequivocal contemporary evidence regarding the desired amount of clearance between cannonball and bore force us to consider the possibility that our categorization does not correspond to contemporary Portuguese practice, Specifically, it is possible that Sacramento's 26 pounders were, in fact, considered 25 pounders in Portuguese arratels.

The reasoning behind this speculation is based on the value of the arratel calculated in the text, on the bore diameter measurements, and on the best available evidence concerning the difference between ball diameter and hole diameter. The most reliable value for this difference, called "windage," is from the late-16th-century Spanish author Luis Collado, virtually the only practical gunner to have written on the subject in the early modern period. Collado, in his Platica manual de artilleria,#44 states that the ball should weigh 10 percent less than the bore; in other words the ball should be 90 percent of the weight of a ball which would fill the bore completely.

Collado's value for windage is generally confirmed by 17th-century English texts, for example, the tables from the 1627 and 1692 editions of. John Smith, A Seaman's Grammar.#45 Collado's rule consistently yields a ball having a diameter 3 percent smaller than the bore, while Michael Lewis' research indicates that the English rule for cannon was 1 percent smaller than the bore and that for culverins was 5 percent smaller.#46

There can be little doubt that Collado's rules for computing windage accurately reflected the accumulated wisdom of the best and most experienced gunners of his day, that is, of the 1570s and 1580s. Better gunners and cannon founders undoubtedly appreciated the theoretical advantages of reduced windage—more muzzle velocity for less gunpowder was seen as the main benefit—an impression strengthened by Collado's discussion of the use of a closely fitting wad to hold the ball in the barrel when shooting downhill.#47 Collado cautions that a reduced powder charge should be used in this case to avoid blowing the gun up; it is apparent from this that he had a clear understanding concerning the relationship between reduced windage, in this case zero, and internal pressure.

It is probable, too, that the bores of Sacramento's cannon were straighter and truer than those of the general run of cannon in Collado's day and that a mid-17th-century cannonball was less likely to be out of round than one of a century earlier. These factors might have permitted a slight reduction in windage from Collado's day, as they undoubtedly did over the long haul.#48 In 1856, the standard windage for an American 42-pound cannon with a 7-inch (17.8 centimeters) bore was .18 inches (.45 centimeters). Application of Collado's formula for the same bore yields a windage of .24 inches (.61 centimeters), or half again as large.

But any presumed increase in manufacturing precision would have been of less consequence, in gross quantitative terms, than the need to provide adequate clearance for the buildup of powder residue in the bore. If our assumptions concerning the level of competence of the early modern cannon founder are correct, these two factors were continually considered and carefully weighed against each other.

All evidence indicates that early modern cannon founders did not use bore diameter as a primary conceptual tool or planning criterion. Rather, they thought in terms of ball weight, always expressed in whole pounds except in the very smallest cannon. Assuming, then, that the intended bore diameter of the six pieces by Lucas Matias Escartim was 5.98 inches (15.2 centimeters), we find that a cannonball of 25 arratels would have weighed 88 percent of the bore "weight" while a 26 arratel ball would have weighed 92 percent of the bore. On balance, the larger value seems the more probable; it is just marginally tighter than Collado's value of three-quarters of a century earlier, which is what we would expect, so our 26 pounders are probably 26 pounders in either system. The difference is a fine one: we are talking about a total difference in windage and ball diameter of only about .04 inches (1 millimeter). Still, we have every reason to believe that the 17th-century cannon founder strived for, and frequently attained, the standards of precision which this implies.

Dr. Guilmartin holds a B.S. from the U.S. Air Force Academy and received his M.A. and Ph.D. from Princeton University where he specialized in early modern European maritime, commercial, and military history. He served as editor of the Air University Review, the professional journal of the U.S. Air Force, from 1979 through 1982. He is presently director of the Space Shuttle History Project under the Center for the History of leadership Institutions at Rice University, working at the NASA Lyndon B. Johnson Space Center in Houston. As an adjunct professor of history at Rice, he offers courses in the history of technology, aerospace history, and military history. He wishes to express his appreciation to the officers and men of the Serviço de Documentação Geral da Marinha, Ministerio da Marinha, Brazil, for their assistance in the project on which this article is based, and in particular to Capitão-de-Mar-e-Guerra (RRm) Max Justo Guedes, director, at whose invitation the research underlying it was performed.

© 1983 by the Society for the History of Technology. All rights reserved.


   #Top                      Home


1 Geoffrey and David Allen, The Guns of Sacramento (London, 1978). For more narrowly technical analyses of specific guns and cannon founding techniques, see Jeremy N. Green, "The Armament from the Batavia," International Journal of Nautical Archaeology and Underwater Exploration (henceforth IJNA) 9 February 1980): 43-51; and John F. Guilmartin, Jr., "The Cannon of the Batavia and the Sacramento: Early Modern Cannon Founding Reconsidered," IJNA, vol. II (June 1982).

2 In addition to periodic reports in the popular press, particularly in Great Britain, work on the Mary Rose has been reported in a series of announcements and articles in Mariner's Mirror (henceforth MM) and IJNA. See, e.g., Peter Whitlock, "King Henry VIII's Mary Rose, 1511-1545," MM 67 (November 1981): 379-82, and "King Henry VIII's Mary Rose-1511-1545, an Outline of the Project and Its Progress," MM 66 (November 1980): 344-48; and Margaret Rule, "The Mary Rose, an Interim Report, IJNA 1(1972): 123-35, and "An Early Gun-Port Lid," MM 62 (May 1976): 184-85.

3 Vanoccio Biringuccio, Pirotechnia, trans. Cyril S. Smith and Martha T. Gnudi (New York, 1942), based on the 1540 Venice edition. For the technical development of bronze artillery before design crystallized along the lines described here, see Heinrich Müller, Deutsche Bronzegeschützrohre 1400-1750 (Leipzig, 1969).

4 Alfred Thayer Mahan, The Influence of Seapower Upon History, 1660-1783 (Boston, 1890); see pp. 287-88 for a particularly clear statement in the context of Mahan's analysis of French commerce raiding strategy during the Seven Years War. For an incisive critique of Mahan's method, see Geoffrey Symcox, The Crisis of French Seapower, 1688-1697, from the Guerre d'Escadre to the Guerre de Course (The Hague, 1974), esp. pp. 227-29.

5 Frank Fox, Great Ships: The Battlefleet of King Charles II (Greenwich, 1980), contains a full treatment of the tactical dominance and lack of endurance and seaworthiness of the giant English three-deckers of the first and second rates and their Dutch and French equivalents. See p. 95 for the tactically decisive role of the largest English ships in the Second Anglo-Dutch War (1664-67) and p. 21 for their unwieldiness. The smaller "rates," which were on occasion employed far from home had two ordnance allowances: an augmented establishment "to be carried only during home waters" (p. 187), which practically meant during the relatively calm months of late spring and summer, and a reduced allowance for peacetime cruising and for extended operations on overseas stations in wartime. It is plain from this that heavy firepower close to home and seakeeping capability on distant stations were incompatible at the time.

6 See Symcox, pp. 109-10, for a discussion of the planning behind the French naval campaign of 1691, and cf. John F. Guilmartin, Jr., Gunpowder and Galleys: Changing Technology and Mediterranean Warfare at Sea in the Sixteenth Century (Cambridge: 1974), pp, 57-84, esp. pp. 76-84.

7 To cite two examples among many, Christopher Lloyd, Atlas of Maritime History (New York, 1975), and Helmut Pemsel, A History of War at Sea (Annapolis, Md. 1977), do not even mention the war by name. The omission in the latter work (originally published in German in 1975 and translated and published in English by the Naval Institute Press with the subtitle, An Atlas and Chronology of Conflict at Sea from Earliest Times to the Present) is particularly revealing.

8 See, e.g,, Michael Lewis, Armada Guns: A Comparative Study of English and Spanish Armaments (London, 1961), p. 167; and Francisco-Felipe Olesa Muñido, La galera en la navigacion y el combate (Barcelona, 1971), I:102-3.

9 See Guilmartin, Gunpowder and Galleys, pp. 277-83, for a full discussion. This phenomenon was well known to ballisticians of the latter half of the 19th century. See, e.g, J.G. Benton, A Course of Instruction in Ordnance and Gunnery ... (New York, 1862). pp. 29, 126-29, and 153: and Thomas Jefferson Rodman, Reports of Experiments on the Properties of Metals for Cannon and the Qualities of Cannon Powder (Boston, 1861), esp. pp. 195-203.

10 Biringuccio (n. 3 above), p. 235, and Guilmartin, Gunpowder and Galleys, pp. 284-91.

11 Thomas Jefferson Rodman, Reports of Experiments on the Strength and Other Properties Metal for Cannon ... (Philadelphia, 1856), pp.286-91. Rodman tested samples of metal cut from the "breech square and "gun head" of a 6-pound howitzer, a relatively short piece.

12 Guilmartin, Gunpowder and Galleys, pp. 284-91, for a summation, and "Os canhoes do Santíssimo Sacramento," Navigator (the journal of the Serviço de Documentação Geral da Marinha, Rio de Janeiro), no. 17 (January-December 1981), pp. 38-39, nn. 14 and 15, and (in English) pp. 78-79, for a detailed analysis of the author's dimensional data and the results of spectrographic and chemical analysis of metal samples extracted from representative cannon.

13 See Guilmartin, Gunpowder and Galleys pp. 170-73 for a general summation. Lewis (n. 8 above), pp. 189-201 thoroughly documents the tendency toward shorter, thinner cannon in England during the late 1500s. My dimensional analysis of 16th- and 17th- century bronze cannon, encompassing a total sample of some 200 guns (mainly Spanish, Portuguese, Venetian, Ottoman, and German), confirms that this tendency was a European and not just an English phenomenon.

14 Examples of this kind of consistency by an exceptional founder include four magnificent 44-pound cannon of battery by the German founder Gregory Leoffler in the Museo del Ejercito, Madrid (Grupo 45 nos. 2826, 2827, 2828, and 3430, cast in 1542, 1546, 1546, and 1543, respectively); despite having been cast at different times, these hardly differ at all in external dimensions and, if the markings are to be believed, differ in weight by only 200 pounds.

15 Carlo Cipolla, Guns, Sails and Empires (New York, 1965), p. 45, n.3.

16 Summarized by Ulysses Pernambucano de Mello e Neto, "O galeão Sacramento." Navigator, no. 13 (June 1976-December 1977), pp. 10-11. The primary published source is Sebastião de Rocha Pita, Historia da America Portuguesa (Lisbon, 1730).

17 Pernambucano. p. 10, from the account in Rocha Pita, pp. 376-80, and "The Shipwreck of the Galleon Sacramento," IJNA 8 (August 1979): 215.

18 Pernambuco, "O galeão Sacramento." p. 9, fig. 1. The ship was lost after missing the channel into the Bay of Todos os Santos in darkness and running aground on a bar at about six o'clock. Then, probably following efforts to lighten the ship which may have compromised her seaworthiness, Sacramento was blown off the bar by a sudden storm from landward shortly before midnight, foundered, and sunk shortly thereafter. Preliminary investigation, based on standard navigational tables, indicates that the moon, at three-quarter phase, rose nearly an hour after sunset in the latitude and longitude of Bahia on the night of the ship's loss. Moonrise had been at sunset on the previous night. It is just possible, therefore, that the ship's loss was due to an incorrect assumption, perhaps by senior supercargo—Sacramento carried the governor general designate of Brazil—that the tricky channel into Bahia harbor could be run by moonlight.

19 Prof. T. Bentley Duncan (Department of History, University of Chicago) to Guilmartin, May 10, 1979, citing correspondence in the Bibliotecha Nacional de Rio de Janeiro, Documentos historicos 9:294-97; Rocha Pita, Historia (1950 Bahia ed.), pp. 238-40; and Frederich Mauro Le Portugal et l'Atlantique au XVIIe siècle (Paris, 1960), p. 86. Duncan's analysis suggests that Sacramento was not bunched before 1651.

20 Information obtained from Capitão-de-Mar-e-Guerra (RRm) Max Justo Guedes, director of the Serviço de Documentação Geral da Marinha, and Vice-Almirante Fernando Ernesto Carneiro Ribeiro, commander of the Second Naval District, encompassing the coastal waters of Bahia, at the time of the wreck's discovery and during the ensuing salvage operations. Positive identification of the galleon was achieved by careful correlation of the dates on the recovered cannon with what was known about vessels involved in major wrecks in Bahian waters. These two men, highly informed students of early modern naval history and technology, were instrumental in the process of identification, a fascinating story in itself, which began with Carneiro's identification of the seven privately recovered cannon in a Salvador salvage yard. Recognizing them as something more than ordinary cannon of Napoleonic vintage, he had them seized as national treasures and set in motion the machinery which ultimately recovered the rest.

21 Pernambucano, "The Shipwreck of the Galleon Sacramento," p. 215.

22 João da Gama Pimentel Barata "Os navios," Historia naval Brasileira, vol.I, tomo I (from Serviço de Documentação Geral da Marinha Rio, de Janeiro, 1975), pp. 80-81.

23 See Symcox (n. 4 above), pp. 57-59.

24 For the impact of strategic and tactical considerations on ship design in a parallel context, see the excellent comparative analysis of French and English design practice by Robert Gardiner, "The First English Frigates," MM, vol. 61 (May 1975). See Guilmartin, Gunpowder and Galleys, pp. 204-12, for the impact of social factors on warship design.

25 Cipolla (p. 56 n. 1) indicates that Portugal initiated large-scale importation of Swedish cast-iron ordnance following the resumption of independence and by 1693 was Sweden's leading customer.

26 Pernambucano, "O galeão Sacramento," pp. 37, 35, photo 52.

27 Rodman, Strength and Other Properties Metal for Cannon (n. 11 above), pp. 152-53, describes a test conducted on two 12-pound howitzers cast eight to ten minutes apart from the same vat of molten metal. They varied in the density of their bronze by 3½ percent.

28 Both English and Spanish documents almost invariably list cannon by barrel weight. For English cannon, for which the record is clearest, the pound avoirdupois was certainly used from the reign of Elizabeth I at the latest (see Ronald E. Zupko, British Weights and Measures: A History from Antiquity to the Seventeenth Century [Madison, Wis. 1977]). For general analysis of units of weight and measure applied to ordnance in the 16th and 17th centuries, see Francisco-Felipe Olesa Muñido, La organización naval de los estados Mediterraneanos y en especial de España durante los siglos XVI y XVII (Madrid, 1968), I:285-88

29 These computations are necessarily approximate and provide only a rough check. They do, however, yield consistent results.

30 Lewis (n. 8 above). p. 219 see also Olesa Muñido, pp. 285-88, for equivalent Mediterranean units.

31 John and Richard Phillips are mentioned in the Calendar of State Papers of August 16. 1588, as purveyors of cannon to the British Crown; George Elkine, who apparently died in 1604, is first mentioned in 1595 (Admiral Sir Terence Lewin. GCB, MVO, DSC, ADC, to Capitão-de-Mar-e-Guerra [RRm] Guedes, June 14, 1977).

32 Based on the appearance of cannon from the Mary Rose, viewed by me in the Museum of the Royal Artillery Institution in the Rotunda at Woolwich, England; the earliest date on these cannon is 1529 (Catalog of the Museum of Artillery, Part I, Ordnance [London, l963], p. 7). This impression is sustained by the appearance of dated cannon in the collections of the Museu Militar, Lisbon; the Museo del Ejercito, Madrid; and the Askeri Musesi ,and Deniz Musesi, Istanbul.

33 Lewis (n. 8 above), p. 201.

34 See B. H. Slicher von Bath, The Agrarian History of Western Europe AD 500-1850 (London, 1963), esp. pp. 113-115, for a concise explanation of the wage and price movements of the 16th and 17th centuries.

35 This is an exquisitely cast 32 pounder some 6 feet (1.82 meters) long, marked CUDEBAT PETRUS GEORGIUS FIGUIEIRA M D C LXXVIII.

36 See Guilmartin, Gunpowder and Galleys, pp. 109-11, for a summary of the evidence in a naval context.

37 Biringuccio, (n. 3 above), pp 246-48.Biringuccio offers several alternate cruzeta designs, including four-part cruzetas and an elaborate structure called a "castle" which vaguely resembles a wrought-iron flowerpot stand and was apparently intended to support the weight of the core as well as centering it. His preference was for a wrought-iron ring with four equally spaced centering rods, as described here.

38 Niccolo Tartglia, Three Bookes of Colloquies concerning the Arte of Shooting in Great and Small Pieces of Artillerie ..., trans. Cyprian Lucar (London. 1588), colloquie 22, p. 41: '....a peece which breakes doth most commonlie breake at the breeche or neere unto the mouth and seldom tymes in the middle...." Tartaglia, a theoretical mathematician and not a gunner, is occasionally vague or inaccurate on points of operational practice, but he is reliable in iris reporting of foundry practice. His opinion is confirmed by examination of surviving examples of burst cannon, notably in the collections of the Museu Militar, Lisbon, and the Deniz Musesi, Istanbul.

39 Cipolla (n. 15 above), p. 55, n. 5, cites a Swedish source to the effect that by 1626 iron cannon were no longer exported from England, and he states that "...according to available figures the [English] fuel [charcoal] crisis seems to have exploded in all its gravity during the l630s'' (p. 63). Cipolla's data are remarkably comprehensive and his analysis convincing.

40 See Kritovolous, History of Mehmed the Conqueror, trans. Charles T. Riggs (Princeton, 1954), pp. 43-46, for an eyewitness account of the casting of such a gun, ca. 1453.

41 Noted by Prof. Joel Shinder, formerly of Fredonia State College, New York, on an Ottoman cannon in the park of the Askeri Musesi. Saint Irene's Church, Istanbul (Shinder to Guilmartin, April 9, 1970).

42 See, e.g., A. R. Hall, Ballistics in the Seventeenth Century (Cambridge, 1952), p. 36, where the early modern gunner's penchant for extreme precision is held up as "a conspicuous example (if the fact that accuracy with balance and ruler will prove very inadequate without the framework of an intellectual system." In fact, the early modern gunner and cannon founder achieved precisely what they set out to do, and accurate mensuration played a key role in their success.

43 Texas Instruments, TI-55 Owners Manual (1977), pp 39-43.

44 Luis Collado, Platica manual de artilleria (Milan, 1592), tractado 3, capitulo 15.

45 Reproduced in E. H. H. Archibald, The Wooden Fighting Ships in the Royal Navy, AD 897-1869 (London, 1968).

46 Lewis (n, 8 above), p. 39.

47 Collado, tractado 3, capitulo 30, fols. 51-52.

48 See, e.g. Rodman, Strength and Other Properties Metal for Cannon (n. 11 above) p.109.

  #Top                      Home