Review Reports - An African Art Re-Discovered: New Revelations on Sword Manufacture in Dahomey

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors present a very thorough investigation into the source of swords from the kingdom of Dahomey. The investigation couples neutron tomography, powder diffraction, and residual stress measurements. The manuscript is very well written and provides a very informative discussion on the true provenance of the Dahomean swords. I have only very minor edits for the manuscript.

I have the following minor comments:

· Page 4, line 97: “Franco-Dahomean wars of 180-1894”, I assume this should be 1890-1894. Please check and correct.

· Table 1: estimated age of the European sword should say 1750-1780.

· Page 5, line 124: the appears to be extra spaces between the and variation.

· Page 7, line 241: “these pores can sometimes by deduced by…” should be “these pores can sometimes be deduced by…”

· Page 8, line 283: delete “the” in “from folding the during forging”.

· Missing references

o Page 8, line 298 and 306.

o Page 15, line 389 and 394.

o Page 16, line 421 and 431.

· Wüstite appears to have an incorrect accent on the u

o Page 18, line 546.

o Page 19, line 553 and 568.

Author Response

Dear Reviewer,

I would like to thank you for your revisions that helped to improve the paper.

Every comment has been considered and individually addressed as reported in the detailed list below.

I hope the revised version is suitable for publication.

Kind regards,

Filomena Salvemini

---------------------------------------

Comments addressed:

C1: Page 4, line 97: “Franco-Dahomean wars of 180-1894”, I assume this should be 1890-1894. Please check and correct.

R1: This has been amended accordingly.

C2: Table 1: estimated age of the European sword should say 1750-1780.

R1: This has been amended accordingly.

C3: Page 5, line 124: the appears to be extra spaces between the and variation.

R3: This has been amended accordingly.

C4: Page 7, line 241: “these pores can sometimes by deduced by…” should be “these pores can sometimes be deduced by…”

R4: This has been amended accordingly.

C5: Page 8, line 283: delete “the” in “from folding the during forging”.

R5: This has been amended accordingly.

C6: Missing references

Page 8, line 298 and 306.
Page 15, line 389 and 394.
Page 16, line 421 and 431.

R6: As explained by the Editor, the issue was inadvertently encountered when the Word document was converted to PDF and uploaded.

C7: Wüstite appears to have an incorrect accent on the u

Page 18, line 546.
Page 19, line 553 and 568.

R7: These has been amended accordingly.

Reviewer 2 Report

Comments and Suggestions for Authors

The paper is very interesting, both in the approach of African objects as in the use of non-invasive multi-methodological techniques to analyze them. The shown results are very impressive and open new and exciting ways to investigate archaeometalurgical objects. The neutron diffraction results are spectacular. The ability to estimate the amount of wustite, fayalite and cementite by a non-destructive technique is very exciting. I wonder if martensite could be differentiated from ferrite, by ND, by the lattice expansion from carbon solution. The amount of porosity detected by tomography in wrought iron is also a total surprise for me.

However, I cannot agree with a very important conclusion the authors have drawn from the results.

“the great probability is that by the mid-to-late 19th century the iron used in Dahomean swords was overwhelmingly African.” Line 669.

This conclusion is based on paragraph starting on line 543:

The diffraction tests also reveal a significant presence of slag in all samples, often a

notable inclusion in bloomery iron in the smelting phase [31]. That is not a definitive in

dicator in itself as some slag can also be present in iron smelted in blast furnaces [32].

Nonetheless, combined with the presence of wṻstite, it is clear this is bloomery iron,

smelted through what is called the ‘direct method’, where the temperature of the furnace

is lower than the melting point of the iron [28]. This direct method was the method used

in all African furnaces prior to the collapse of indigenous iron production in the early 20th

century.

Why this is inappropriate? Even wrought iron produced in Northen Europe in 1860, by the puddling process, as a part of the indirect process, had a large amount of slag inclusion, containing wustite dendrites in its interior.

A citation from a Buchwald´s paper (one of the cited authors) indicates

Puddled irons have more slag inclusions than do the fined Scandinavian irons. The slags are present as voluminous, wüstite-rich stringers surrounded by a host of small slag droplets. The ferritic state of the iron corresponds well to the abundance of wüstite, which often constitutes more than 60 vol.% of the slags.

https://doi.org/10.1016/S1044-5803(97)00105-8

Puddled iron is a technique used to transform pig iron, the product of blast furnace, into wrought iron, through oxidation of the molten pig iron, stirring the semi-solid material and taking it to a rolling mill.

In fact, it is not easy to differentiate if that type of slag inclusions come from direct or indirect processes. Dillmann and L´Heritier proposed a method, based on the content of P2O5 in the slag inclusions, even then with a lot of uncertainty.

P. Dillmann, M. L’Héritier, Slag inclusion analyses for studying ferrous alloys employed in French medieval buildings: supply of materials and diffusion of smelting processes. J. Archaeol. Sci. 34, 1810–1823 (2007). https://doi.org/10.1016/j.jas.2006.12.022

As a conclusion from my side: The amount of slag identified by Neutron diffraction indicates that the swords have not been made using the technologies used in Europe after 1860, when the Bessemer and the open hearth processes were developed to produce a much cleaner liquid steel.

Appart from that, some details must be corrected:

Line 96: need to correct the date 180?

Line 104: This sword was brought to France in 1890 ?

Page 410: fayalite as a result of added SiO2?

Fayalite is one of the major constituents of slag inclusions, and forms through reaction of the wüstite with the silica (e.g. sand) or other silicates (e.g. clays) added as flux [26] during smelting.

One must not forget that silica is the most common contaminant of iron ores.

Line 549: why it is said that the carbon levels are below two percent? Sword 7, the only one where cementite was detected, had 0,3%C. So I can be said that carbon from swords from 1 to 6 have is at least below 0,3%.

Line 540: “The presence of wṻstite, an oxide that forms when oxygen particles are reduced in the smelting process, is indicative of incomplete reduction, and only occurs when iron is smelted at a low temperature (26).”

This is not correct. It must be a misreading of Buschwald´s paper. Of course there are no oxygen particles, but more important than that, the presence of wustite is not necessarily indicative of incomplete reduction, as all puddled iron from the 19 century also show a significant amount of wustite. It may come from the oxidation of iron during refining.

Line 562:

“In the process, two separate materials were produced – pig iron, a type of iron almost devoid of carbon, and molten cast iron, very high in carbon, which could be refined further into steel by decarburisation. The need for this additional refining process is why this was called the ‘indirect’ method [23, 33].”

Pig iron is an alloy of iron plus 4%C that can only be used for cast pieces. Pig iron may be refined to produce low carbon wrought iron. If one wants to be more detailed, pig iron may be produced with low silicon, and called white iron, or with 2%Si, when it is called gray iron. White iron was used as a raw material for the refining, whereas gray iron was used for castings.

The paper is very good, the results very impressive, it must be published. Only the main conclusion is not as strong as the authors would like it to be.

Author Response

Dear Reviewer,

I would like to thank you for your revisions that helped to improve the paper.

Every comment has been considered and individually addressed as reported in the detailed list below.

I hope the revised version is suitable for publication.

Kind regards,

Filomena Salvemini

---------------------------------------

Comments addressed:

C1: The paper is very interesting, both in the approach of African objects as in the use of non-invasive multi-methodological techniques to analyze them. The shown results are very impressive and open new and exciting ways to investigate archaeometalurgical objects. The neutron diffraction results are spectacular. The ability to estimate the amount of wustite, fayalite and cementite by a non-destructive technique is very exciting. I wonder if martensite could be differentiated from ferrite, by ND, by the lattice expansion from carbon solution. The amount of porosity detected by tomography in wrought iron is also a total surprise for me.

However, I cannot agree with a very important conclusion the authors have drawn from the results.

“the great probability is that by the mid-to-late 19th century the iron used in Dahomean swords was overwhelmingly African.” Line 669.

This conclusion is based on paragraph starting on line 543:

The diffraction tests also reveal a significant presence of slag in all samples, often a

notable inclusion in bloomery iron in the smelting phase [31]. That is not a definitive in

dicator in itself as some slag can also be present in iron smelted in blast furnaces [32].

Nonetheless, combined with the presence of wṻstite, it is clear this is bloomery iron,

smelted through what is called the ‘direct method’, where the temperature of the furnace

is lower than the melting point of the iron [28]. This direct method was the method used

in all African furnaces prior to the collapse of indigenous iron production in the early 20th

century.

A citation from a Buchwald´s paper (one of the cited authors) indicates

https://doi.org/10.1016/S1044-5803(97)00105-8

Dillmann, M. L’Héritier, Slag inclusion analyses for studying ferrous alloys employed in French medieval buildings: supply of materials and diffusion of smelting processes. J. Archaeol. Sci. 34, 1810–1823 (2007). https://doi.org/10.1016/j.jas.2006.12.022

R1: Concerning martensite: resolving martensite and ferrite properly is extremely difficult and unreasonable for our kind of data measured in a medium resolution (WOMBAT). High resolution neutron diffraction data would be required (ECHIDNA), but even this is difficult (if not impossible) due to presence of multiple other phases, while looking for extremely small effects (e.g. asymmetric broadening of ferrite peaks) related to the presence of martensite.

In view of the comments received on the conclusions, title and abstract have been revised, while most of the section 9 has been modified. The most salient part added is:” Historical reports aren’t illuminating on the provenance of iron in Dahomey either. Up until now it has been assumed that the iron used in Dahomey was imported from Europe, even during the 19th century, when trade relations had changed after the 1807 Slave Trade Act abolishing human trafficking internationally. British visitors to Dahomey – the British Consul Richard Burton, in the 1860s, and J. A. Skertchly in the early 1870s – both attested to the fact that Dahomey didn’t produce its own iron [46,47]. Both were aware of huge iron smelting centers in the interior but stated that most iron in Abomey came from European imports on the coast. Skertchly specifically said that the iron came from England, which would suggest it was ‘voyage iron’, exported from Liverpool. However, he provides no actual evidence. It’s also clear that Skertchly based many of his observations on what Burton had said in his visit a decade before, and there is no evidence that Burton’s statement on this was anything more than an assumption.

According to Skertchly: “Very little (iron) is bought from the tribes to the north, and this is exclusively used for fetiche purposes” [47]. Fetiche was a derogatory word used to describe anything related to the vodun religion and derives from a Portuguese word that means ‘witchcraft’. It is well documented that West Africans believed European iron to be ritually inadequate – the term ‘deaf’ has been used to connote its insensate quality in all things metaphysical or spiritual [48]. So, it is highly unlikely it would ever have been used in religious ritual if local iron was available.

This is relevant because all the swords examined here were ritual objects. Swords 1, 2 and 3 were objects of royal regalia in a kingdom where there were no neat divisions between religion and the state. Symbols on these swords not only reference Dahomean royalty, but also the iron god Gu [4, 6]. Furthermore, swords 4 and 5 were undoubtedly used in vodun ritual in one form or another, for they are both sacrificial swords. Both swords have a small cutting edge on the bottom edge of the blade and are simply not functional for fighting, only decapitation – either of small animals, in the case of sword 4, or larger animals or humans, in the case of the longer and heavier sword 5 [4]. Furthermore, sword 6 has no practical utility at all, for it has no sharp cutting edge and with the figural hand at the blade extremity it could not be used for stabbing or thrusting. It’s only conceivably function is as a cult or ritual object.

Magico-religious rites and observances relating to both the smelting and forging of iron were taken extremely seriously all-over precolonial Africa, even when it came to iron for ordinary domestic or agrarian use [49]. This was also the case in precolonial Dahomey where propitiation of the iron god Gu was universal. However, the magical qualities of iron were even more important when it came to objects connected to religious practice. The importance of this is far too well documented to be ignored. While prior to the puddling method imported iron was no doubt genuinely inferior, even pejoratively called ‘rotten iron’ in certain places in Central West Africa (40), after 1784 that was likely not the case. However, local iron was still preferred – not necessarily because it was a better or harder material, but simply because it had the right spiritual credentials [48].

There were other uses that seem to have been deemed more appropriate for European iron – including making bullets for the muskets and Dane guns still widely in use, which utilized iron ball pellets. The practice of making bullets from bars of imported iron was well recorded among the neighboring Yoruba by British observers in the Yoruba wars of that time [50], It was also observed by the Scottish adventurer John Duncan in 1849 when in Abomey, Dahomey’s capital [51]. It is likely that European iron was used for making agrarian tools. Indeed, Evans & Rydén have made a direct correlation between increased iron imports and periods of agricultural growth [33]. However, given the data on vodun ritual and local veneration of iron as an aspect of the god Gu, it seems unlikely that European iron would have been considered appropriate for these swords.

On the surface, the thing that seems to have convinced Skertchly, and perhaps Burton too, that Dahomean iron was imported was the mere fact that it came in the form of bars and rods. Skertchly certainly witnessed this [47]. However, it is now known that iron bars were also manufactured in Africa. Kriger demonstrates how African bloomery iron was sometimes produced in bar iron units in Central West Africa in the 19th century, where they functioned as currency – one of many iron currency forms. These bars weren’t always fashioned at the smelting stage – sometimes smiths hammered them into bars from rougher iron bloom material [29]. In fact, far closer to Dahomey, David has drawn on early colonial accounts to show that the Sukur smiths in the Mandara Mountains, on the boundaries of Nigeria and Cameroon, fashioned bloomery iron into a semi-finished bar iron product, which was exported in large quantities to the Lake Chad region. Importantly, this was happening before colonial markets had even opened up in that region [52, 53]. While it’s unlikely that either of these were a direct source of iron for Dahomey, both offer a precedent for African bar iron in pre-colonial times, independent of European iron imports from the sea.

As for rods, African iron was fashioned into rods in the form of ‘Kissi pennies’ – bloomery iron from the Sierra Leone and Liberia region (54). African iron currencies existed in a bewildering variety of forms and it’s certainly not impossible that there were other iron rods in circulation as currency in Africa. On the other hand, it’s not known whether rods were ever imported from England. Indeed, the prevailing studies suggest voyage iron was provided in bar form only, to strict specifications with very clear stamps. After all, iron bars also functioned as a form of specie [33]. However, current historical studies are far too wide ranging, and more studies are needed to differentiate voyage iron in the 18th century from imported iron in the 19th century, after the passing of the Slave Trade Act of 1807, as well as the Equipment Act of 1839. The abolition of slave trading would surely have changed the entire trading dynamic, but so far studies on voyage iron have not examined that. It is also not clear whether the introduction of a puddling process would have changed the pattern of export – after all, puddling is an English invention, even though it was also adopted in Sweden.

In conclusion, the scientific data do not currently offer indisputable evidence of the provenance of the iron. However, at the very least, the neutron data here show that the iron used in these swords was just as likely to be African iron as imported iron from Europe or Brazil. That represents a significant step from current assumptions that most iron in 19th century Dahomey was imported from Europe. In fact, those assumptions don’t really stand up well to scrutiny, especially as more and more studies bring the sophistication of Africa’s pre-colonial iron industries into focus. However, they also can’t be completely dismissed yet. Future isotype analyses could help make new breakthroughs, with the help of destructive methods, including those initiated by Dillman & L’Heritier [44]. However, more data on African iron would be needed.”

C2: Line 96: need to correct the date 180?

R2: This has been amended.

C3: Line 104: This sword was brought to France in 1890 ?

R3: This has been amended.

C4: Page 410: fayalite as a result of added SiO2?

One must not forget that silica is the most common contaminant of iron ores.

R4: The sentence has been modified as follow: “Fayalite is one of the major constituents of slag inclusions, and forms through reaction of the wüstite with the silica, present as common contaminant in the iron ore, or other silicates (e.g. clays) added as flux [26] during smelting.”

C5: Line 549: why it is said that the carbon levels are below two percent? Sword 7, the only one where cementite was detected, had 0,3%C. So I can be said that carbon from swords from 1 to 6 have is at least below 0,3%.

R5: The sentence has been modified as follow:” Occam’s razor, the law that posits that the simplest solution is usually the best, might favour the conclusion that this is African iron. The fact is – the data suggest that the iron in all six African swords has the characteristics of iron smelted in the bloomery furnaces that were used in iron production all over pre-colonial Africa [43]. Bloomery furnaces em-ploy what is called the ‘direct method’, where the temperature of the furnace is lower than the melting point of the iron, which might explain the presence of wüstite [28]. By con-trast, the iron in the European sword included for comparative purposes (sword 7), could not have been formed by the same process. Rather than wüstite, it has cementite – an iron carbide that illustrates that this material was smelted at very high temperatures. As this is steel, it could only have been smelted by the ‘indirect method’, where the temperature of the furnace was higher than the melting point of the ore.. The difference in the diffraction data between the six African swords and the European sword could not be more pro-nounced. Furthermore, the diffraction tests also reveal a significant amount of slag in all six Dahomean samples. Slag inclusions are a notable feature of bloomery iron, and much less so of more modern refining processes [31]. Nonetheless, despite all this preliminary evi-dence, the findings on smelting are less cut and dry than the findings on forging.

C6: Line 540: “The presence of wṻstite, an oxide that forms when oxygen particles are reduced in the smelting process, is indicative of incomplete reduction, and only occurs when iron is smelted at a low temperature (26).”

R6: The paragraph has been modified as follow:” According to the diffraction data, all six swords are composed of ferrite (a-iron), and there are no signs of any other mineral microstructures, such as cementite or pearlite, which are associated with high carbon iron. Indeed, the data show that all six swords can be defined as wrought iron. As already discussed, the presence of wüstite can be indica-tive of incomplete reduction during the smelting process, and frequently forms when iron is smelted at a low temperature (26)”.

C7: Line 562:

The paper is very good, the results very impressive, it must be published. Only the main conclusion is not as strong as the authors would like it to be.

R7: The paragraph has been modified as follow:” In the process, two separate materials were produced – pig iron, a type of iron almost de-void of carbon, and molten cast iron, very high in carbon, which could be refined further into steel by decarburisation. The steel in the European sample (sword 7) was transformed from molten cast iron in this way, which is why this was called the ‘indirect’ method, since an additional step [23, 33].

The pig iron could also be refined into wrought iron through a process called pud-dling, first by oxidising the high silicon content to form white cast iron, and then by re-ducing the carbon by stirring the metal in a coal burning furnace. Developed in England in 1784, puddling was soon adopted by other major iron and steel exporters, including Swe-den, and only supplanted by other technologies towards the end of the 19th century [26].

Both bloomery iron and puddled iron are both defined as wrought iron because of their low carbon content. Both have wüstite, though it may not have occurred for the same reason. In bloomery iron it may be a result of incomplete reduction. However, in puddled iron it may result through oxidation. Both irons also have significant slag inclusions. Though there may be less slag in puddling iron, for slag is scraped off the surface in the process, it is difficult to establish this as a universal characteristic. So, it can be difficult to distinguish between bloomery iron and wrought iron – in essence, they are the same mate-rial, even though they are produced by different methods.

Some research breakthroughs have been made in distinguishing wrought iron pro-duced through the direct method from wrought iron produced through the indirect meth-od – in other words, bloomery iron from puddled iron [44]. This has involved examining slag inclusions to determine what fuel was used during smelting. However, importantly, such research is dependent on destructive techniques, and so is beyond the scope of this study, which has sought to preserve these swords.

The only other mineral present in the six Dahomean swords is fayalite, which shows that sand or silica was likely used as a flux at some point in the manufacture. Of course, silica is a common contaminant in iron. Around the world, blacksmiths use flux to im-prove welding, partly by removing contaminants at the join, but also by lowering the tem-perature needed for welds to occur on the surface of the metal. In Africa, such fluxes were also used during smelting bloomery iron to lower the melting point of iron ore [28].

Reviewer 3 Report

Comments and Suggestions for Authors

The authors must consider minor corrections:

- page 6, lines 175-176: It would be appropriate to include the numerical values of the ranges of the d-Spacing and Peak Resolution used parameters

- page 8, lines 265 and 266: the authors mention that “Figure 4.1 revealed the metal to be free of any detectable defect”, but in this figure there is a long, almost horizontal crack at the tip of the sheet, so it is suggested to modify the statement.

- page 8, lines 270 to 276: The presence of texture engravings on the surface of the swords at the tip, especially on the sword 3, is missing on Figure 4.3. In fact, authors don´t mention superficial details such as engravings on the samples that are visible on tomography images.

- page 8, line 298: erase “Error! Reference source not found.” and write “Feret shape 3D, sample 6”.

- pages 8, line 306: erase “Error! Reference source not found.” and write “(Figure 5.7)”.

- pages 11 and 12, Section 5. Neutron Residual Stress Analysis: Figure 7 is not mentioned or referenced in this section, please add a mention to Figure 7.

- page 12, lines 328 to 338: It is recommended to mark the positions of cross-sections 1 and 2 measured in Figure 4.1, with a dotted red line to locate them and have reference to the sites that were analyzed.

- page 15, line 388: clarify whether they refer to Table 4 and not Table 3.

-page 15, line 389: the idea of the statement from "Error! Reference source not found _ and ranging within 1-2 wt%"; in addition, Figure 9 does not graphically present the wÜstite phase as mentioned. Correct and clarify the idea.

- pages 15, line 394: erase “Error! Reference source not found.” and write “Figure 8”.

- page 15, line 397: the authors write "2 different groups" in the text, but in line 415 and the list of groups, there were 3 different groups and not 2 different groups... correct.

- page 16, line 431: verify what the authors mean in this line, it is confusing.

- page 16, line 437: change "Figure 2" to "Figure 10" because the numbering is incorrect.

Author Response

Dear Reviewer,

I would like to thank you for your revisions that helped to improve the paper.

Every comment has been considered and individually addressed as reported in the detailed list below.

I hope the revised version is suitable for publication.

Kind regards,

Filomena Salvemini

---------------------------------------

C1: page 6, lines 175-176: It would be appropriate to include the numerical values of the ranges of the d-Spacing and Peak Resolution used parameters.

R1: The requested info was added.

C2: page 8, lines 265 and 266: the authors mention that “Figure 4.1 revealed the metal to be free of any detectable defect”, but in this figure there is a long, almost horizontal crack at the tip of the sheet, so it is suggested to modify the statement.

R2: The crack visible at the tip is related to an indentation of the chiseling decoration. This sentence has been added for clarity: “Here, as well as in swords 3 and 4, since the tomographic cross sections are near the surface, the visible gaps in the metal are due to decorative engravings”.

C3: page 8, lines 270 to 276: The presence of texture engravings on the surface of the swords at the tip, especially on the sword 3, is missing on Figure 4.3. In fact, authors don´t mention superficial details such as engravings on the samples that are visible on tomography images.

R3: This has been addressed in the previous comment.

C4: page 8, line 298: erase “Error! Reference source not found.” and write “Feret shape 3D, sample 6”.

R4: As already explained by the Editor, the issue was inadvertently encountered when the Word document was converted to PDF and uploaded.

C5: pages 8, line 306: erase “Error! Reference source not found.” and write “(Figure 5.7)”.

R5: As already explained by the Editor, the issue was inadvertently encountered when the Word document was converted to PDF and uploaded.

C6: pages 11 and 12, Section 5. Neutron Residual Stress Analysis: Figure 7 is not mentioned or referenced in this section, please add a mention to Figure 7.

R6: The reference to Figure 7 has been added on the first sentence of section 5 on line 324: “The neutron diffraction residual stress analysis was conducted on all Dahomean blades with the exception of sword 4 (Figure 7).”

C7: page 12, lines 328 to 338: It is recommended to mark the positions of cross-sections 1 and 2 measured in Figure 4.1, with a dotted red line to locate them and have reference to the sites that were analyzed.

R7: This has been added together with a comment in Figure 7 caption: “The positions of the two cross-sections are mapped as red and blue dotted lines on the photographic image of the blade at the top”.

C8: page 15, line 388: clarify whether they refer to Table 4 and not Table 3.

R8: This has been amended.

C9: page 15, line 389: the idea of the statement from "Error! Reference source not found _ and ranging within 1-2 wt%"; in addition, Figure 9 does not graphically present the wÜstite phase as mentioned. Correct and clarify the idea.

R9: It seems there is a confusion/misreading here. The wurstite phase in the Figure 9 is referred by its chemical name, FeO, not by its mineral name. But the sentence makes it clear referring two ways.

C10: pages 15, line 394: erase “Error! Reference source not found.” and write “Figure 8”.

R11: As already explained by the Editor, the issue was inadvertently encountered when the Word document was converted to PDF and uploaded.

C11: page 15, line 397: the authors write "2 different groups" in the text, but in line 415 and the list of groups, there were 3 different groups and not 2 different groups... correct.

R11: This has been amended.

C12: page 16, line 431: verify what the authors mean in this line, it is confusing.

R12: The sentence has been modified as follow: “However, these results (Figure 10) demonstrate variable distribution and abundance of FeO, which is concurred with the observed variability of the residual stress distribution or defect distribution”.

C13: page 16, line 437: change "Figure 2" to "Figure 10" because the numbering is incorrect.

R13: This has been amended.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

It is great to have the opportunity to read again such an interesting paper, with so much more information added.

Now, If you don´t want to be criticized by topics that are not central to your paper, I suggest you a few minor changes.

the phrase starting on line 551, although right in its conclusion, it is using arguments that are inappropriate:

By contrast, 551 the iron in the European sword included for comparative purposes (sword 7), could not 552 have been formed by the same process. Rather than wüstite, it has cementite – an iron 553 carbide that illustrates that this material was smelted at very high temperatures. As this is 554 steel, it could only have been smelted by the ‘indirect method’, where the temperature of 555 the furnace was higher than the melting point of the ore.

1. The presence of cementite is not related to higher temperature, but to more reducing atmosphere.

2. Steel was not only produced by indirect method. People were making steel centuries before the blast furnace.

3. It is not right to assume that the temperature of the furnace had to be higher than the melting point of the ore. In fact, in the blast furnace, as the pig iron reached 4%C, its melting temperature lowered to 1150°C, much below the melting temperature of hematite.

phrase in line 570 has to be changed,

pig iron, 570 a type of iron almost devoid of carbon, and molten cast iron, very high in carbon, which 571 could be refined further into steel by decarburization.

Pig iron IS NOT a type of iron almost devoid of carbon. Pig iron has around 4% carbon, which is a lot.

Pig Iron IS cast iron.

Those are just details. The paper is very good.

Author Response

Dear Reviewer,

Thank you for your in-depth revision.

The comments have been addressed as reported in the detailed list below.

I hope this version is suitable for publication.

Kind regards,

Filomena

-----------------------------------------------------------------

C1: the phrase starting on line 551, although right in its conclusion, it is using arguments that are inappropriate:

The presence of cementite is not related to higher temperature, but to more reducing atmosphere.
Steel was not only produced by indirect method. People were making steel centuries before the blast furnace.
It is not right to assume that the temperature of the furnace had to be higher than the melting point of the ore. In fact, in the blast furnace, as the pig iron reached 4%C, its melting temperature lowered to 1150°C, much below the melting temperature of hematite.

R1: the sentence has been modified as follow: “Rather than wüstite, it has cementite – an iron carbide that illustrates that this material was created in a very effective reducing atmosphere, where oxidation was arrested. This indicates very high temperatures, such as those attained in a blast furnace.”

C2: phrase in line 570 has to be changed,

pig iron, 570 a type of iron almost devoid of carbon, and molten cast iron, very high in carbon, which 571 could be refined further into steel by decarburization.

Pig iron IS NOT a type of iron almost devoid of carbon. Pig iron has around 4% carbon, which is a lot.

Pig Iron IS cast iron.

R2: The sentence has been modified as follow: From this process, two separate cast iron materials were produced – grey iron, characterized by graphite flakes, and another cast iron component without graphite flakes and a higher concentration of cementite, often described as ‘white iron’. As cementite can act as a hardening agent, strengthening the steel, the white iron could be refined further into steel after decarburization. The steel in the European sample (sword 7) was transformed from white molten cast iron in this way, which is why this was called the ‘indirect’ method, since there was an additional step [23, 33].