THE INFRARED SPECTRA OF AMBER AND THE IDENTlFICATION OF BALTIC AMBER

BY C. BECK, E. WILBUR, S. MERET, D. KOSSOVE AM) K. KERMANI

Department of Chemistry, Vassar College, Poughkeepsie, New York

INTRODUCI'ION Until well into the second half of the nineteenth century amber found at

archaeological sites anywhere outside the Baltic region was considered an import from the North. Several Southern deposits of amber or amber-like fossil resins were known, including those of Sicily. but they were not seriously considered as a source of raw material for ornaments of early Mediterranean cultures. Literary sources supported this view. Amber was brought to Greece by Phoenician traders;' we do not know whence, but since they broughmt tin from the British Isles, they could well have brought amber from not much farther away, viz. from the German and Danish coast and coastal islands of the North Sea. The Roman amber trade with Northern Europe is well attested by the Elder Pliny.2

Von Sadowski first established the land routes by which amber reached the Black Sea and the Mediterranean by an ingenious reduction of Ptolemy's longitudes to modern values combined with a thorough knowledge of the topology of Eastern Europe.3 Notable work on the amber trade has since been done by De Navarro4 and by Spekke.5

While amber routes were neatly laid out across Europe it may have seemed churlish to raise the question whether all the amber found in the South need at all times have been transported along these paths. Yet that perfectly reasonable question was asked by Capellini6 at the Jnternational Congress of Anthropology in Stockholm in 1874. His suggestion that Terramare and Villanovan artifacts could have been fashioned from local resins rather than of Baltic amber could not be backed by evidence, but in the absence of evidence to the contrary it could not be dismissed. In 1911, Baudoin7 raised the same question about the neolithic amber artifacts of Southern France. As even late in the nineteenth century native amber was so plentiful in the Department des Basses-Alpes that the peasants there used it for fuel.8 his doubts are equally reasonable.

The problem of determining the provenance of amber by analysis appealed to chemists. Otto Helm, apothecary in Danzig, was the first to make a systematic investigation of ambers of various geographical origin with a view of finding characteristics which would allow him to identify archaeological finds.

I t had been known since the middle of the 16th century that Baltic amber contains a small amount of a characteristic organic acid which sublimes when the resin is subjected to destructive distillation. This compound, succinic acid, was in fact prepared on a commercial scale by this method. Helm9 improved the quanti- tative determination of succinic acid by working out a wet method involving base hydrolysis followed by precipitation of the barium salt which he either weighed directly or used to liberate the free acid which crystallises as the monohydrate. He found the succinic acid lowest in clear Baltic amber (3.2 to 4.5%). higher in the white opaque variety called bone amber (5.5 to 7.804). and highest in the weather- ing crust (8.2%). Helm further measured melting points (of the whole resin as well as of portions extracted by a number of solvents), densities, solubilities, and elemental composition.i0 By such means, but mainly on the basis of succinic acid

A R C H A E O M E T R Y 97

content,il Helm identified a number of important amber finds as having come from the Baltic littoral, among them the beads which Schliemann had found in the shaft graves at Mycenae.12

Helm later13 realized that non-Baltic European fossil resins from Galicia, Hungary, Rumania, the Bukowina and even Sicily contain succinic acid in varying amounts which may be as high as those in Baltic succinite. He dealt with these difficulties in a manner which suggests that his theory was too dear to him to permit evidence to weaken it. Dahms who himself compiled extensive comparative data of European fossil resins14 has said that both the method and the results of the succinic acid analyses were inadequate to support all of Helm’s conclusions.~5

After Helm’s death, Olshausen carried out additional analyses of amber,16 and collected all the earlier work on geological and archaeological amber samples in a table which he did not live to publish, but which was appended to a critical review of the problem by LaBaume in 1935.17

The table lists 49 analyses of geological samples, of which 29 are of Italian or Sicilian origin. Among the latter, three contain from 2 to 5.7% succinic acid. Olshausen also found more than five per cent of succinic acid in raw amber from Figueira, Portugal, the Isle d’Aix on the West Coast of France, and Lemberg in Galicia (now L‘vov, USSR). These notable exceptions throw rather more than reasonable doubt on the conclusions which have been drawn from the succinic acid content of 59 artifacts which Olshausen culled from the Iiterature or made himself.

Even this quite summary review of earlier chemical work on amber in the service of archaeology shows that for all their merit the methods employed and the results obtained fail to settle the problem of the provenance of amber artifacts conclusively. It is clear that future work could profit from past experience in several ways:

(1) The number of analyses of geological samples (i.e. of samples of known origin) must be sufficiently large to give a broad statistical basis to the interpretation of results from archaeological samples.

To make this possible, the method of analysis should require little time and skill, so that a great many analyses may be carried out by workers who have not been trained in highly specialized analytical techniques.

(3) As in all problems confronting the archaeometrist, the analysis should be either non-destructive or else use so little material that sampling will not impair the scientific and aesthetic value of the artifact.

(4) If possible, the analytical method should give not one but several parameters which have potential diagnostic value for distinguishing the bewildering number of varieties of amber and related fossil resins. The most evident disadvan- tage of all earlier work lies in that every single property of amber has proved to fall on a scale on which there is overlap among species.

These considerations led us to investigate the infrared spectra of whole amber in the solid state. The sample is prepared by grinding two milligrams of amber with a hundredfold excess of pure potassium bromide in a dental amalgamator and pressing the powder into a clear pellet in an evacuated die at 80,000 psi. After trials with infrared spectrophotometers of high resolving power (Perkin-Elmer Models 21 and 221), we found the capabilities of the less elaborate Perkin-Elmer Model 237 grating instrument with a range of 4,000 to 625 cm-1 (2.5 to 16 p ) equal to the task. The large number of absorption bands which amber shows in this region

(2)

98 A R C H A E O M E T R Y

offers a multiplicity of parameters from a single experiment which can be per- formed in 20 minutes, or in very much less time if several samples are prepared simultaneously.

We have briefly reported before on the result of the first 120 spectra.18 Since then Savkevich and Shaksig have published a very detailed discussion of the infra- red spectra of ten samples of Baltic amber.

The interpretation of the infrared spectra of fossil resins calls for restraint. The complexity and the heterogeneity of the material makes it difficult to assign bands to highly specific functional groups or structural types. Few bands are sharp, and even sharp bands are rarely reproducible to within better than 20 to 50 cm-1 on samples taken from the same specimen. Still greater shifts can be found when com- paring spectra of samples from different, though mineralogically and geographically identical, specimens. Shoulders abound, indicating the presence of many closely related compounds or structura: features. The intensities of absorption bands show even greater variation than do their wavenumbers. The superposition of 24 spectra of samples from a single specimen of Baltic amber which appears quite homo- geneous under low-power magnification will illustrale this point (Figure 1).

73- 75;

77-81;

85 -100.

8 9 10 1 1 1 2 CI

Geologigches Staatsinstitut Hamburg

FIG. 1. Comparison of 24 Spectra of Baltic Amber Sample B-1.

A R C H A E O M E T R Y 99

Our assignments, and our criticism of assignments by others,'g are based on the evaluation of nearly 600 spectra and take into account the limits of reproducibility we have observed. A typical spectrum of Baltic amber over ,the entire range is shown in Figure 4, Spectrum 95.

OXYGEN-HYDROGEN BONDS The oxygen-hydrogen deformation vibrations are not readily distinguishable

from carbon-oxygen stretching bands even in simple alcohols.20 but since we have found no diagnostic use for them, these uncertainties are happily irrelevant to our purpose.

The oxygen-hydrogen stretching band is always broad and extends from 3100 to 3700 cm-I (2.7 to 3.2 p ) , i.e. over the entire range of free and associated hydroxyl groups. In our spectra it is usually one of the two strongest bands with about the same intensity as the carbon-hydrogen stretching absorption. We have however found that the intensity of the oxygen-hydrogen stretching band and also of the deformation band at 1640 cm-1 (6.1 p ) increases dramatically with grinding time (Figure 2). This clearly shows that new oxygen-hydrogen bonds are formed as new surfaces of resin are exposed in the grinding process, either by thorough adsorption of atmospheric moisture or through oxidation by atmospheric oxygen. There are no significant changes which would indicate epoxides (865 cm-1. 1165 cm-'. and 1265 cm-') or peroxides (830-890 cm-'), but the latter band is weak and difficult to identify, especially in large molecules where skeletal frequencies

3.0 3.5

I l l l l l l t l l Spectrum 78

ground 10 mir F

i 3; 5

1111111111 Spectrum 7 9

around 15 min

i 3;s

- L L d u u h Spectrum 8 0

round 2 0 rnin.

3:O 3; 5

J . . W Spectrum 81

B a l t i c Amber Sample 8-1; Geologischer Staats inr t i tu l Homburg

FIG. 2. OH and CH Stretching in Amber.

100 A R C H A E O M E T R Y

intervene or in alcohols which may themselves absorb in that range. On the evidence of the parallel changes of the band at 1640 cm-' (6.1 p; see below), the increase in the oxygen-hydrogen stretching band may be due to the adsorption of water, but if so, it is held tenaciously, for we have not been able to remove it by drying for several days, with or without phosphorus pentoxide. Savkevich and Shaks have found that the oxygen-hydrogen siretching band increases in their pellets even after storage in a desicator for several days.19

Since a grinding time of more than ten minutes is necessary to produce the uniformly small particle size required for reproducible spectra by the potassium bromide technique, we have sought to standardize rather than to eliminate the effect: all our samples are ground for exactly 15 minutes. Large differences in the oxygen-hydrogen stretching absorption could then be used to distinguish resins by their relative avidity for adsorbing water if not by their original composition. Thus far we have not found such differences which can be related to sample type.

CARBON-HYDROGEN BONDS The strongest absorption in the carbon-hydrogen stretching range is clearly due

to the methylene doublet at 2926 cm-1 and 2853 t 10 cm-1 (3.42 and 3.50, p) and the methyl doublet a: 2962 cm-1 and 2872 f. 10 cm-1 (3.38 and 3.48 p), of which the former seems to predominate and the latter appears as more or less obvious shoulders. Only in a small group of North American fossil resins which are characterised by a large proportion of aromatic constituents are the carbon- hydrogen stretching frequencies sharply separated.

The presence of a large number of methyl groups in Baltic amber also seems to be indicated by the fact that the carbon-hydrogen deformation bands at 1450 1 20 cm-1 (6.9 p; due to methylene and symmetrical methyl bending) and at 1375 f 5 cm-l (7.25 p; due to asymmetrical bending) have approximately equal intensities.

Tertiary hydrogens absorb Loo weakly to be identifiable in the carbon-hydrogen stretching region in most of our spectra, but a commonly occurring shoulder near 1340 cm-1 (7.46 p) may be due to their deformation mode.

All these bands are in keeping with the expectations for terpenoid compounds and afford no means for distinction in our spectra of whole amber. On these same expectations the consistent though broad band between 1050 and 950 cm-1 (9.5 and 10.5 p) could be assigned tentatively to the cyclohexane frequencies identified by Marrison21 in this region.

More critical information can be derived from absorption frequencies associated with hydrogen atoms attached to carbon-carbon double bonds. Savkevich and Shakslg have claimed to recognize olefinic unsaturation of the type CR, R, = CH,, CR, R, = CHR,. and RCH = CH,, as well as aromatic stretching frequencies, but the evidence decidedly does not warrant such detailed assignments. The small but distinct absorption at 3095 cm-1 (3.23 p) is the only olefinic carbon-hydrogen stretch- ing frequency which can be identified unmistakably because it is well out of the range of all other strong bands in this region. It may equally well arise from the groupings RCH = CH, or CR, R, = CH,, but we cannot agree with the Russian authors that both are separately recognizable. While the associated carbon-carbon stretching frequency near 1650 cm-1 (6.05 p) is obscured by water absorption, a sharp band at 885 cm-1 (11.3 p) is almost certainly due to the out-of-plane carbon- hydrogen bending of the terminal olefin. Its position favours the structure

A R C H A E O M E T R Y 101

CR, R, = CH, over the alternative RCH = CH,. In light of the fact that Schmid22 and co-workers have isolated agathaline (1, 2, 5-trimethylnaphthalene) as a dehydro- genation product from Baltic amber, this suggests a diterpene of the agathic acid type with an exocyclic double bond.

The intensity of the 11.3 band is variable, probably as a result of oxidative changes which will be discussed below, but it survives at least as a distinct shoulder in all the samples of Baltic amber which we have examined. The band is of further interest because it allows distinctions to be drawn among other fossil resins. Of the Baltic non-succinite varieties, only gedanite shows it, lbut glessite, krantzite, and beckerite do not. Of other European fossil resins neudorfite, euosmite, and rumanite show it, but duxite, ajkaite, walchowite, schraufite, and the French “pseudo- succinite” do not. Sicilian amber sometimes, but not always, absorbs near 11.3 p (Figure 3). Several resins classed as copalites, from such diverse locations as England, Columbia, and the West Indies all have an 11.3 p band, as does the ambrite of New Zealand. In North America ambrosine from South Carolina absorbs at 11.3 p (Figure 4, Spectrum 328), while retinite and the so-called “burmite” of New Jersey do not. Bucaramangite from Columibia, several fossil resins from Japan, and genuine burmite all fail to absorb in this region.

N e w York N a t u r a l His tory Museum I

BIG

R 7 2 9 9 S m i t h s o n i a n Ins t i tu t ion

I

Smi thson ign I nrti tu t ion -

d -2

C o l u m b i a Univers i ty

FIG. 3. Infrared Spectra of Sicilian Amber.

102 A R C H A E O M E T R Y

\flw-- Geol. St a a t s in st i tu t

Y Hamburg

Spectrum 9 5

M a g o t h y River, Md.

S m i t h r a n i a n R 7 2 8 7

S p e c t r u m 3 1 5

C e d a r Lake, Mani toba

Smi thron ian 9 7 080

V Spect rum 3 1 9

c A M B R O S l N E

Charleston, S.C.

Smithronian R 7 3 1 7

Spectrum 3 2 8

1 4 6 I ? 10 I l I 2 1 .L

FIG. 4.

A R C H A E O M E T R Y 103

CARBON-OXYGEN DOUBLE BONDS All fossil resins show strong carbonyl absorption bands between 1770 and

1695 cm-1 (5.65 and 5.90 p). Very faint changes in slope on the edges of this band, too faint in most cases to be called shoulders, indicate that this is an unresolved composite of several types of carbonyl groups including both ketones and esters. Since the absorption never extends significantly beyond 1685 cm-1 (5.93, p ) , the presence of a, p-unsaturated ketones appears to be ruled out; but both open-chain and five- or six-membered alicyclic ketones are within the range of absorption.

The band is too broad to permit inferences about the nature of the ester groups present, but some distinctions will be made on the basis of the carbon-oxygen single bond absorption in the next section.

CARBON-OXYGEN SINGLE BONDS The carbon-oxygen stretching vibrations between 1250 and 1100 cm-1 (8.0 and

9.0 p) define the spectral range in which the most important and diagnostically useful distinction between Baltic amber and non-Baltic European amber can be made.

Baltic amber and only Baltic amber shows a highly characteristic absorption pattern in this range. It consists of a broad and typically perfectly horizontal shoulder between 1250 and 1175 cm-1 (8.0 and 8.5 p ) which is followed by a sharp absorption peak which reaches its maximum intensity just below 1150 cm-1 (8.7 p), after which absorption rapidly diminishes. There is an automatic filter change in our instrument at 1150 cm-1 (8.7 p) which causes a small but discernible jump of the recording pen. This mark is visible in the illustrations shown here, and in all samples of Baltic amber it coincides or immediately follows the sudden decrease in absorption by the sample.

We have found this absorption pattern to be unique to Baltic amber. It is apparent in the infrared spectra published by Moenke23 and by Savkevich and Shaks,lg but none of these authors have recognized it as a distinguishing mark of Baltic amber. There can be 110 doubt that the band is due to the carbon-oxygen single bond stretching of saturated aliphatic esters, rL, ,&Unsaturated esters and aromatic esters absorb below 1150 cm-1 (above 8.7 p); neither seem to be present in Baltic amber. One cannot help but recognise a certain similarity of the ester absorption of Baltic amber with that of diethyl s~ccinate ,~" which raises the amus- ing possibility that the most striking feature in the spectra of Baltic amber is nothing more than a rapid ard convenient short-cut to the more laborious succinic acid analyses which Helm applied to the same end, viz. the distinction of Baltic amber from other fossil resin. We are now preparing a variety of succinate esters and recording their spectra. The results will be published at a later time.

Yet while appreciable amounts of succinic acid have been found in non-Baltic European amber,l7 none of about 350 spectra of non-Baltic European amber which we have taken shows an absorption pattern between 1250 and 1100 cm-I (8.0 and 9.0 p) which matches that of our 250 spectra of Baltic amber.

The ester absorption bands of the non-Baltic European fossil resins are extra- ordinarily variable, and we have not yet succeeded in relating them to geographical origin. The spectra of four samples of Sicilian amber are shown in Figure 3.

The qualification non-Baltic European amber is necessary because we have found ester absorption in several (but not all) samples of North American fossil

104 A R C H A E O M E T R Y

1 1 1 I

resins which is so similar to that of Baltic amber as to be indistinguishable from it (Figure 4). The paleobotanical inferences of this similarity, which happily does not affect applications to European archaeology, has been discussed elsewhere.25

1 I I i _ - - - -

DETERIORATION OF AMBER While we have demonstrated the essential reproducibility of spectra of a given

specimen (Figure l), a discussion of the infrared absorption of amber cannot ignore the considerable variation among spectra of different specimens from the same location.

In non-Baltic ambers these variations are so great that they stand in the way of identifying any of them from spectral evidence alone. The present indications are that some non-Baltic fossil resins, e.g. simetite from Sicily and chemawinite from Cedar Lake, Manitoba, include resins of divergent botanical origin, and that it will be to the identification of source trees, rather than geographical areas, that future efforts should be directed. To what extent such botanical assignations can or will lead to clearer distinctions among geographically or mineralogically defined types of fossil resins remains to be seen.25

In Baltic amber, too, there is considerable variation among spectra, and since it affects, inter alia, the absorption band ,between 1250 and 1100 cm-1 (8.0 and 9.0 p ) which is the prime diagnostic feature of Baltic amber, it must be considered here.

We have noted above that ideally the ester absorption of Baltic amber includes a broad, horizontal shoulder followed by a peak at 1150 cm-1 (8.7 p ) , as shown in spectrum 74 (Figure 5). While this pattern is realised in most of our samples of Baltic amber, the shoulder shows a negative slope in many others, e.g. in spectrum 179 (Figure 5). We have sought to explain this change and to support the explana- tion empirically.

The absorption in the critical range is due to the stretching of carbon-oxygen single bonds. Now exposure to 'the atmosphere will inevitably lead to the forma- tion of new carbon-oxygen bonds, and these will absorb at the same approximate

A R C H A E O M E T R Y 105

wavelengths, thus changing the absorption pattern between 1250 and 1100 cm-1 (8.0 and 9.0 p). Moreover, the new carbon-oxygen bonds must inevitably be formed at the expense of carbon-carbon and carbon-hydrogen bonds some of which should correspondingly decrease in intensity or disappear altogether. Acmong the most vulnerable groups in Baltic amber should be the terminal methylene groups which are responsilble for the narrow absorption band at 885 em-i (11.3 p) which has been discussed above in the section on carbon-hydrogen bonds. One might therefore predict that exposure to the atmosphere should lower the intensity of the band at 885 cm-1 (11.3 p) and simultaneously change the slope of the shoulder between 1250 and 1150 cm-l(8.0 and 8.7 p ) .

That these two changes go hand in hand is indeed evident from a comparison of spectra 74 and 179 (Figure 5). Unfortunately we know nothing of the history of these two geological samples except that the apparently more thoroughly oxidised one from Heligoland was found very much farther from home than the more perfectly preserved one from the Baltic coast. We therefore looked for some un- questionably Baltic amber with a known history of exposure and found it in the City Museum of Rostock, East Germany, Spectrum 196 (Figure 6) is of a sample excavated from the Slavic fortifications at Rostock-Dierkow. Archaeologically, it should be of the order of a thousand years old, but we cannot say how porous the soil in which it was buried may have been or for how long it may have been kept in the open air before it was buried again. Spectrum 194 (Figure 6) is of a piece of amber found in a megalithic grave at Ziesendorf (Mecklenburg). Its archaeo- logical age is several times as great as that of the Rostock-Dierkow sample, and for all of this time it was more or less exposed to air.

1 9 4

v

f

!

Megalithic Grave (Neol i thic Age)

Ziesendorf

Slavic Fort i f icat ion

Rortock- Dierkow

Museum dcr Stadt Rostock, Mecklcnburg, Germany (DDR)

FIG. 6.

106 A R C H A E O M E T R Y

The expected change of the slope between 1250 and 1150 cm-1 (8.0 and 8.7 p) is evident in both spectra, as is the reduction of absorption at 885 cm-1 (11.3 p) to a mere shoulder. Both are more pronounced in the older sample.

Some amber artifacts of low intrinsic value gave us further opportunities to observe the progressive deterioration of Baltic amber under the influence of air.

Figure 7 shows duplicate spectra of seven plaques of amber which were inlaid in a cabinet of Sicilian workmanship and of uncertain but relatively recent date. Since Baltic amber and even copal has long been substituted in Sicilian shops for the rarer and more expensive native product, these inlays (now kept in the Depart- ment of Geology at Columbia University, New York City) made a convenient test case. The regions between 8 and 9 p shown in the figure leave no doubt that the amber is indeed of Baltic origin. It is gratifying to be able ,to add that those samples which show the greatest change in slope in the figure (samples marked C 1 1 , C 12,

c 1 1

-.J 2 4 1

ui 2 4 2

c 1 2 C 1 3 I

2 4 3 I 2 4 5

c 1 4

ci 2 4 7

L/ 2 4 8

C 1 7 c 1 6

2 4 8 I 2 5 3

2 5 0 I 2 5 4

il 2 5 6

FIG. 7. Inlay from Sicilian Cabinet, Columbia University, Dept. of Geology. 8-9 I"

and C 17) also show a nearly complete loss of the terminal methylene peak at 885 cm-1 (11.3 p) to a mere shoulder, while a weak but clear absorption peak remains at this wavelength in the other four.

Figure 8 shows 28 spectra of beads from a necklace of Etruscan origin which is now in the Classical Museum at Vassar College. Its precise findspot is not known since it was bought from a dealer. but a pendant representing a frontal view of a crudely carved face connects it stylistically with well-authenticated amber carvings of the same type in the Zwinger at Dresden and in the Metropolitan Museum of Art in New York.26

Amber from chamber tombs as they are common in the Etruscan domain will predictably present severe problems of oxidation. The beads of the necklace, as well as the pendant, showed deep cracks and a dark reddish-brown colour, both usual consequences of oxidative degradation. As expected, the spectra are correspondingly affected, but most of them leave no doubt that the amber is of Baltic origin. Only

A R C H A E O M E T R Y 107 ~~~

the spectra of bead 4 (spectrum 387), bead 7 (spectrum 388), and bead 47 (spectrum 370) might possibly be taken for Sicilian amber (compare Figure 3). In all the spectra of this necklace, the band at 885 cm-I (11.3 p ) was reduced to a barely visible shoulder.

.It thus appears possible to settle the origin of amber artifacts in terms of the large distinction Baltic amber versus non-Baltic European amber even in the face of extensive oxidative damage, although it may be necessary to resort to multiple sampling and to avoid surface samples. We think it well to run at least two spectra of each specimen initially, and to run more if the first leave room for doubt.

We should prefer to remove the oxidised material rather than to resign ourselves to its presence. Mechanical separation has not proved rewarding, but we have had

3 9 7

B e a d 1 1

400

Bead 2 2

Beod 30 3 70

Bead 4 7

FIG. 8. Infrared Spectra of Etruscan Amber Necklace between 7.5 and 9.0 Microns; (Vassar College Classical Museum)

some success with solvent extraction. Since the air oxidation of amber increases its solubility (probably by reducing the average molecular weight) the exhaustive extraction with various organic solvents leaves a residue which gives a spectrum more nearly like that of unoxidised amber. There is an attendant advantage in that spectra can at the same time be taken of the dissolved material. In these solutions absorption bands of some functional groups (e.g. the carbonyl groups of esters and of ketones, respectively) are more sharply separated than they are in the spectra of whole amber in the solid state. We have hopes that these solution spectra will come to be useful in distinguishing not only between Baltic and non-Baltic amber, but also between the many varieties of non-Baltic amber.

108 A R C H A E O M E T R Y

CONTAMINATION OF AMBER ARTIFACTS This report cannot be complete without a brief note about the contamination

of amber artifacts which can seriously interfere with the interpretation of their infrared spectra.

One common impurity in archaeological specimens is earthy material which may have penetrated deeply into a cracked surface. Some of the inorganic anions found in such impurities, such as silicates, sulphates, and phosphates, absorb so strongly between 1250 and 1000 cm-1 (8 and 10 p) that any appreciable amount of them may obliterate the ester absorption of the amber itself quite hopelessly.

The second kind of impurity is deliberate: many amber artifacts in museu’ms (particularly those we have seen in Italy) have been treated with paraffin, beeswax, or synthetic polymer solutions in order to consolidate them and to improve their superficial appearance. Paraffin is of least concern since it contains only carbon- carbon and carbon-hydrogen bonds; but beeswax and many synthetic polymers, like polyvinyl acetate, contain ester linkages which obscure the diagnostically im- portant region from 1250 to 1100 cm-1 (8.0 to 9.0 p).

We have therefore found it necessary to inspect every sample closely under a low-power microscope before choosing the sample for the infrared spectrum, and to separate natural and deliberate contaminants mechanically from the sample.

ACKNOWLEDGEMENTS This work was supported by ,the American Philosophical Society for the

Advancement of Useful Knowledge. We are indebted for samples of amber to Dr. W. Hantzschel of the Geologisches Staatsinstitut in Hamburg, Dr. Brian Mason of the New York Museum of Natural History, Dr. George S. Switzer of the U.S. National Museum (Smithsonian Institution), Prof. Paul F. Kerr of the Department of Geology, Columbia University, New York, Mr. Harry Wiistemann of the City Museum of Rostock, Germany (DDR), and Prof. Inez S. Ryberg of the Department of Classics, Vassar College. We also thank Prof. Richard C. Lord of the Massachusetts Institute of Technology, Cambridge, Mass., for a most helpful discussion of some aspects of this work.

REFERENCES

1 Homer, Odyssey IV, 73; XV, 460; XVIII, 2%. Cf. W. Helbig, “Das Homerische Epos aus

2 Pliny, Naturalis Historia, Book XXXVII. 3 J. N. von Sadowski, “Die Handelsstrassen der Griechen und RBmer’, transl. by A. Kohn,

Jena, 1877. Reprinted Amsterdam, 1%3. 4 I. M. De Navarro : Geogr. J . , 66,481-507 (1925), with map. 5 A. Spekke, “The Ancient Amber Routes and the Geographical Discovery of the Eastern

6 G. Capellini, Cong. internat. d’Anthrop. Stockholm 1874 (pwbl. 1826), pp. 803 ff. 7 M. Baudoin, Revue du Bas-Poitou, 1911, 180. 8 L. Bonnemere, Bull. SOC. Anthrop. Paris [3] 9, 122 (1886) 9 0. Helm, Arch. Pharm. [3] 11, 229 (1877).

10 0. Helm, Schrift. Naturf. Ges. Danzig, N.F. 5, (3) 9 (1882).

den Denkmalern erlautert”, Leipzig, 1887.

Baltic”, Stockholm, 1957, maps.

A R C H A E 0 M E T R Y 109

11 For reasons of his own, Helm fell back on the rather crude method of liberating succinic acid from amber by dry distillation in his work on Schliemann’s finds. Much later R. Jones (Schrift. Phys.-Oekon. Gex Kijnigsberg, 49, 351 (1908) used wet hydrolysis with sodium

methoxide on Mycenaean amber finds and calculated the succinic acid content without isolating the acid or its salts.

12 0. Helm, Schrift. Naturf. Ges. Dmzig , N.F. 6 , (2) 2 (1885). 13 0. Helm, ibid. 7, (4) 189 (1891) and 10, (4) 37 (1902). 14 P. Dahms, ibid. 10, (2-3) 243 (1901). 15 P. Dahms, ibid. 12, (2) lf1908). 16 0. Olshausen and F. Rathgen, Ztschr. Ethnologie, 1904, 153. 17 W. LaBaume, Schrift. Naturf. Ges. Danzig, N.F. 20, (1) 5 (1935). 18 C. W. Beck, E. Wilbur, and S. Meret, Nature, 201, 256 (1964). 19 S. S. Savkevich and U. A. Shaks, Zhur. Priklad. Khim., 37,930 and 1120 (1964). 20 L. J. Bellamy, “The Infrared Spectra of Complex Molecules”, 1958, p. 108. All assignments

21 L. W. Marrison, 1. Chem. Snc., 1951, 1614. 22 L. Schmid, Monatsh., 72, 290 and 31 1 (1939). 23 H. Moenke, Chem. Erde, 21, 239 (1961). 24 Standard Spectrum No. 3001, Sadtler Research Laboratory, Philadelphia, Penn. 25 J. Langenheim and C. W. Beck, Science, 149, 52 (1965). 26 E. Richardson, Yale University, private communication

of infrared bands in this paper are cited from Bellamy.