Volume 66 1957 > Volume 66, No. 1 > Some chemical, physical and structural properties of moa egg shells, by Cyril Tyler, p 110-130
SOME CHEMICAL, PHYSICAL AND STRUCTURAL PROPERTIES OF MOA EGG SHELLS
OVER THE LAST few years Tyler and his colleagues at Reading, 1 have developed a number of methods to help them in their studies of hen egg shells and it is now becoming clear that these methods may be of interest when studying the egg shells of wild birds and even the shells of extinct birds. Dr. Skinner, of the Otago Museum, very kindly supplied the author with six samples of moa shell and this paper describes the results obtained with these. The author is not qualified to assess these results from any point of view other than that specifically related to shell structure and composition, but it is hoped that the results will encourage others who are interested in the problem of the moa to use these and similar methods.
For this reason the paper must be regarded as suggesting methods which might prove of value when examining moa shells and not as a contribution to the classification of moas.
GENERAL DESCRIPTION OF EGG SHELLS.
In order to understand the methods employed it is perhaps necessary to remark briefly upon the question of egg shell structure.
Figure 1 shows a radial section through an egg shell and this may be regarded as a somewhat conventionalised, but generally true picture for most, if not all, avian egg shells. On the inside are two membranes- i
Photographs of the outer (a-f) and inner (g-l) (mammillary) surfaces of the six samples of shell. Each photograph represents one square millimeter of shell.
- ii Page is blank- 111
and attached firmly to these are the knob-like mammillae which in turn change imperceptibly into the so-called spongy layer, which, in actual fact, is very compact. On the outside of the shell there is the cuticle. The mammillary and spongy layers consist almost entirely of calcium carbonate, with small quantities of magnesium, sodium, potassium, phosphate and citrate, and in addition there is an organic matrix. A concentration of organic material is found as a core near the base of each mammilla. Starting from gaps between clusters of mammillae there are channels, or pores, which pass through the spongy layer and open into a depression in the shell surface. In most birds' eggs these pores have single channels, but in the rhea they may sometimes be branched to give two or three openings, whilst in the ostrich they are multi-branched and give a whole series of openings clustered together in one depression.
In 1949 Oliver briefly summarised the work done on moa egg shells. This appears to consist of three papers all published before 1900. First, came a paper by Nathusius (1870), who described the mammillae and spongy layer of the shell, and gave some excellent drawings of vertical sections through pore systems. These systems show that the elongated depressions in the shell surface may have as many as three mouths opening into them, but all arising from one pore. The rounded depressions on the shell, however, represent pores with single mouths. Nathusius concluded that the moa egg shell is most nearly like the rhea egg shell. The second paper came from Hutton (1871) and it was unfortunate that he had not seen the paper by Nathusius which gave better and more detailed descriptions. However, Hutton also likened the moa egg shell most nearly to the rhea. Finally a paper by Liver-sidge (1880) gave some values for the chemical analysis of the moa egg shell. The fragments were from several different shells and most of them were described as weathered.
Oliver (1949) discusses linear pores which he contrasts with punctures, but he does not go into further detail about pore structure. As far as the author is aware no detailed study of chemical, physical and structural characteristics is available.
Dr. Skinner provided six samples of moa shell, each presumably homogeneous and coming from particular sites. The samples were as follows and will be refererd to throughout the paper by the serial numbers given to them by us.
A number of quite large pieces of shell was available in each sample. In some samples these different pieces were somewhat variable, - 112 therefore it has only been possible to give a very general description which quite clearly would not necessarily apply to a particular piece. In addition the pieces photographed, and shown in Figure 2 have been selected to emphasize certain differences.
1201. The shell was buff coloured, with little staining on either surface and no deposits. Pore depressions occurred chiefly as slots, but on some pieces these were full of debris (Fig. 2a). Mammillae were quite clearly marked and there was a hole in the end of most of them. Some fragments were etched on both surfaces and where this occurred the holes in the ends of the mammillae were much enlarged (Fig. 2g).
1202. The shell was greyish-white with brown stains and most of the pieces were very badly eroded giving an irregular surface. Extreme erosion coincided with large round pits in the shell (Fig. 2b), often filled with debris. Mammillae were usually clear cut but the centres were eaten out (Fig. 2h).
1203. The shell was buff coloured, and there was no sign of erosion or deposits, except that some of the pore slots had a slight deposit. The pore slots had two or more pore mouths leading into them (Fig. 2c). The mammillae were frequently, but not always, clearly separated and many showed the hole in the end very distinctly (Fig. 2i).
1204. The shell was buff coloured and had glistening deposits over part of both surfaces (Fig. 2d), such deposits made the pore slots much shallower, or covered them completely. The mammillae were small and only a few showed the hole in the end (Fig. 2j).
1205. The shell was pale grey in colour and badly eroded in parts. Long shallow troughs occurred but in some cases these were filled almost to the surface with a deposit (Fig. 2e). There was little indication of the mammillae (Fig. 2k). Some of the pores were filled with a dark deposit.
1206. The shell was buff coloured or grey and had a smooth surface. In some cases there were deep slots into which pore mouths opened, in others round dimples (Fig. 2f). The mammillae were difficult to see in some pieces (Fig. 2l), but in others were quite clear cut with holes in the centre of each one. Most of the pores of this sample were filled with a yellow to orange material, and embedded in this were darker grains.
Hutton (1871) reported a thickness of about 1.7 mm., and Oliver (1949) stated that he had never observed a thickness greater than 2.0 mm. In the present work measurements were made using a micrometer screw guage and reading to the nearest 0.01 mm. Five measurements were made on each of ten pieces of shell from each sample, and, for comparison, similar measurements were made on samples of rhea and ostrich shell from which the membranes had been removed by the method of Tyler and Geake (1953a).
For each sample the mean thickness of each piece was calculated and from the five measurements used to obtain this mean a value for the coefficient of variation was also calculated. This coefficient of variation - 113 measures variations in thickness within any one piece and the mean of the ten coefficients of variation for the ten pieces of one sample (mean c.v. within pieces) gives an average value for variations within pieces of a sample. The mean thicknesses for the ten pieces were then used to calculate a general mean for the sample and also the coefficient of variation. This coefficient of variation gives the variation in thickness between different pieces of the same sample (c.v. between pieces). The main results are given in Table 1.
Mean shell thickness (mm.) of different pieces and different samples and coefficients of variation.
At the outset sample 1202 will be ignored since it is so badly eroded. Considering the other samples it is evident that their general means are not very different from each other, and certainly not sufficiently to suggest any species differences. In comparison with the rhea and the ostrich all samples, except 1204, gave a much greater c.v. between pieces, in fact greater than would be expected within one shell. This suggests that some external factor has been at work, probably erosion, caused by weathering or by chemical and/or microbiological events in the soil, or perhaps by man-made abrasion. Sample 1204 has a lower c.v. between pieces than the rhea shell and its value is little greater than that for ostrich shell. This probably indicates very little erosion, unless erosion was equal over all pieces; a most unlikely event. It is of great interest to observe that the thickest piece measured of each sample gave a range from 1.34 to 1.46 mm. and, apart from 1206, all are between 1.40 and 1.46. Since this group includes the uneroded 1204 it may be that the thickest pieces of the other samples also represent uneroded material. If this is so, then there is excellent agreement between the original thicknesses of the five samples and the variation between the samples is less than would be expected within other species or even between egg shells from individual birds.
For each sample the most variable piece showed a c.v. of between 2.29 and 2.96, a figure just above the rhea value, but much above the ostrich value. The least variable piece gave values ranging from 0.00 to 0.78 with rhea at 0.52 and ostrich at 0.23. Apparently erosion does not create a very large c.v. within pieces in these cases, hence it must be fairly uniform over a given piece. This was supported by the fact - 114 that there was no relationship between the thickness of any one piece of shell and the variability in thickness over that piece (c.v. within piece). If erosion was not uniform over a piece it is probable that the greater erosion would produce a greater variation along with a greater degree of general thinning.
The case of sample 1202 is quite exceptional. The thickest piece is much thinner than any of the others, and the general mean is low. The c.v. between pieces is the highest of all the samples, but not much higher than 1205. Similarly, the individual pieces show a great deal of variation. Examination of this sample indicates all this quite clearly and it is doubtful if measurements of the type under discussion are worthwhile on samples as badly eroded as this.
In order to obtain more information about the question of erosion the shells were examined in greater detail. It was observed by Nathusius (1870) and by Hutton (1871) that two layers could be distinguished in the moa shell and the writer found a similar result in many, but not all, pieces examined. The two layers seem to correspond to the mammillary layer (inner) and the spongy layer (outer). These are always present, but not always easy to see sufficiently well to make accurate measurements. The best examples are those in which the mammillary layer appears crystalline and the spongy layer is covered with a white amorphous powder. In such pieces there is a clear line of demarcation between the two layers.
Only a few examples of shell showing these layers clearly and therefore capable of measurement were available, and these were measured at a number of points on the edges, using a micrometer eyepiece on the microscope. At the same time a description of the outer and inner surface of the piece was written down. Table 2 shows the results. Since 1204 appears, from evidence quoted above, to be the least eroded of all samples, it will be considered first. Between each of the four pieces
Thickness (mm.) of the spongy and mammillary layers of shell, with notes on the condition of the shell surfaces.
there is practically no variability and the mean overall thickness is 1.40 mm. for these pieces; equal to the mean given in Table 1. These four pieces were covered by ten readings in all and the general mean thickness is divided into an outer layer of 0.97 mm. and an inner one of 0.43 mm. The c.v. for the whole shell is 1.18, for the outer layer 1.08 and for the mammillary layer 3.84. The variation is thus very small. Again, the evidence suggests that 1204 is a sample which has suffered little erosion on either surface and this is once more supported by the appearance of the outer surface showing pores depressions, varying from round to very elongated and filled in many cases with debris. The inner surface shows clear cut mammillae with few or no holes in the ends and is glistening in appearance. The proportion of the two layers is 69.7% outer and 30.3% inner which corresponds well with Hutton's (1871) two-thirds and one-third.
With this shell as a standard, the others may now be considered. Pieces 1201/6 and 1201/9 showed litle signs of erosion on either surface and the mean values are 0.95, 0.44, 1.39 mm., for the outer layer, inner layer and total shell thickness respectively, giving 68.3% outer and 31.7% inner layer. These values are remarkably close to the values for 1204 and suggest similarities in these uneroded pieces of shell. The piece 1201/2 is only 1.26 mm. thick and it is obvious from the data in the table that erosion has occurred on the outer, but not on the inner, surface, whilst 1201/4 is 1.02 mm. thick and has lost material from both surfaces. Such findings suggest therefore that care must be taken, when examining pieces of shell, to judge the weight to be given to a value for thickness. Only pieces with no sign of erosion on either surface should be used. A further very important point is the question of elongated and rounded pore depressions. In sample 1204 pore depressions varied from round to much elongated ones. The uneroded pieces of 1201 showed the same picture, but 1201/2, showed smaller and rounder depressions and 1201/4 showed even smaller round pore depressions. This might have suggested pieces of shell from another species, and overall thickness supports this, but examination of all the data gives strong support to the idea that erosion of the outer surface has given rise to cross sections of the pore channels much lower down their length, and hence more nearly round in cross-section.
Sample 1205 gave results fitting in well with the previous concepts of erosion, with 1205/4 showing an extreme degree of mammillary erosion. The values for the two layers in the apparently most normal piece (1205/5) are 71.4% outer and 28.6% inner layer, figures again not far removed from 1201 and 1204.
It was, unfortunately, only possible to obtain two values from one piece of 1206. These gave 0.85 mm. (72.3%) for the outer, 0.33 mm. (27.6%) for the inner and 1.18 mm. for total thickness. This suggests that erosion had probably occurred, however, the pore depressions are clean and many are greatly elongated, whilst the mammillary layer shows no erosion and few holes in the ends of the mammillae, but it may be that they are so eroded as to be partly worn away. The mean c.v. within pieces (Table 1) was the lowest of all values, hence further evidence, in the case of this shell, must be considered.- 116
Sample 1202, as expected gave results of great variability, some pieces having lost much of the outer layer, some of the inner layer and some losing parts of both. It should be noted, however, that in most pieces of 1202 the outer layer could be subdivided into two, the outermost one tending to appear very powdery compared with the other.
It was unfortunate that insufficient material from sample 1203 was available for this test.
Recently Tyler (1956) has described methods which, by using a plastic embedding technique, give casts of pore channels and also of the surface of the mammillary layer.
In one method the piece of shell is embedded in plastic and then the plastic is carefully filed away to expose the outer surface of the piece, but not the edges or the inner surface. Concentrated hydrochloric acid is then used to dissolve away the shell and the plastic is unattacked. When the hydrochloric acid has completely dissolved the shell, there remains a cavity in the plastic. On the bottom of this depression is the cast of the mammillary layer, and, projecting from it, like stalagmites, are the casts of the pores. These may be examined in situ, then removed with dissecting needles and mounted on cavity slides. In some samples a white jelly-like material is left in the cavity and sticking to the pores, but it is fairly readily removed. On drying it becomes a white powder and then it is more difficult to remove.
A second method is to file the plastic in such a way that only an edge (radial section) of the shell is exposed. Treatment with acid for a few seconds will then etch this surface and in many shells the etching is differential, i.e., different layers of the shell dissolve away at different rates, and these can be readily distinguished under the stereoscopic microscope. 2 The explanation of this is not yet certain but it is probably partly owing to the different extent to which the original liquid plastic monomer penetrates the various parts of the shell. Then the greater the amount of polymerised plastic there is present, the more slowly will the acid attack the calcium carbonate and also the greater will be the plastic residue. A further point about this method is that if a pore is in, or near, the surface of the exposed shell face then the acid will etch the material away from around it and it will stand out clearly, running from its origin in the mammillary spaces to the outer surface where it usually opens in a depression.
The methods have been applied to all samples of the moa shell, but it is proposed merely to give a general summary of the results.
Both methods have given satisfactory results and it is clear that there are a number of variations of pore shape within any one shell, but that all shells have given the range of variation. Figure 3 shows drawings of a selection of these traced from actual photographs of pore casts dissected off after using method 1. Firstly there are pores of fairly uniform bore and with very little widening at the mouth; there- 117
FIG. 3.- 118
Drawings, traced from photographs, of pore casts, showing a selection of different types. a, c and e are varieties of pores with double mouths, b is a single pore, d is a single and double together opening into a common pore depression and f is a collar of plastic removed from a pore cast of 1202. Actual measurements are given of the pore length and, in the case of f, its diameter.
are other single pores which tend to show a flattening at the mouth, thus giving an oval shaped pore mouth. Secondly there are pores which branch to form two mouths and others give three mouths. In this the moa shell resembles the rhea, which also gives three kinds of pores. Nathusius (1870) has already commented on this situation, but there is one additional point to make, namely, that pores often form into groups with a common depression to serve all the pore mouths for that group and Fig. 3d shows a single and a double mouthed pore opening into a common depression. This grouping of pores and pore mouths gives rise to the long narrow depressions usually orientated in one direction, observed in moa shells, but the length of these depressions will vary in any one shell from relatively short ones housing one pore mouth to long ones housing as many as four pore mouths and, in one case, five were noted. The proportion of each kind might be of some diagnostic value. In Fig. 4 all the combinations of pores giving rise to the varying numbers of pore mouths per depression up to a maximum of four, have been shown diagrammatically. It is possible, if sufficient moulds were made, that all combinations would be found, and, in fact all these have been observed except c/i and d/i.
Figure 5, a, b, c and d show some actual photographs of pore casts in position in radial sections. The one in (b) is stained by material originally present in the pore channel. The one in (c) is incomplete but the cast was whole when the section was first made.
The samples used all gave many more casts of aborted pores than is usual, but this may have been caused by the fact that so many pore depressions are blocked by debris and hence prevent the entry of the liquid plastic to the fullest extent.
It has already been mentioned that pore holes per square cm. and their mode of grouping might offer some diagnostic evidence. Many pieces of shell were therefore examined under the stereoscopic microscope and counts made. Then, knowing the area of the field examined, it was possible to multiply up the values to give readings per square cm. Table 3 gives the results, with a count on rhea shell for comparison. Only values for three moa shells could be obtained in sufficient numbers to give a reasonable mean and in the case of sample 1201 it was necessary to dip each piece in concentrated hydrochloric acid for about
Total number of pore depressions and pore mouths and distribution of pore mouths per depression. Mean values per sq. cm. and percentage distribution in parenthesis.
FIG. 4.- 120
Diagramatic representations of all combinations of pores giving rise to varying numbers of pore mouths per depression up to a maximum of four. Those types marked with an asterisk have actually been observed.
two seconds in order to get rid of the debris in the pore depression, thus making a count possible. Because of the shortage of results and the special treatment necesary for sample 1201 the value of these figures should not be overstressed; nevertheless, there are one or two points of interest. The total number of pore mouths per square cm. is variable and so is the total number of depressions; furthermore the number of depressions showing single, twin or triplet holes shows considerable variation. It is known that different hens tend to give different pore counts per square cm. and that there are species differences, so that the results found here are not surprising. Clearly the rhea gives a far greater percentage of single pore depressions.
A few measurements were also made on the length of the pore depressions in relation to the number of pore holes and the distance from centre to centre of pore holes in a depression, but the results were not very reliable. However, with more readings per shell the values might be of interest.
When radial sections were etched, samples 1201, 1203, 1204, gave no indication of differential etching, thus showing no particular layered structure as, for example, Tyler (1956) found with many birds, including the rhea. In Fig. 5, 1204 is shown at (a) and 1203 at (d), and both are equally etched at all levels; in each case the etching only occupied a few seconds. Samples 1205 and 1206 however, gave much more etching of the mammillary than the spongy layer, Fig. 5b shows sample 1205 after about five minutes etching. The spongy layer left behind a residue of a fibrous nature, the fibres appearing to run parallel with the surface of the shell. As before, sample 1202 was quite different from all the others. Acid attack from the surface (method 1) was slower but was ultimately complete; this left a white felted layer which could easily be lifted off. It was made up of many subsidiary layers which tended to separate on drying. Its underside showed lumps also covered with a fine white material and through these projected plastic pore casts; often the cast had a collar of white material which could be removed as a disc with a hole through it (Fig. 3f). A piece of shell etched on its radial section gave confirmatory evidence in that an outer layer and a thinner layer roughly two-thirds of the way down the shell proved to be resistant (Fig. 5c). It would therefore appear that erosion has considerably increased the size and depth of the pore depressions, giving rise to relatively round depressions on the surface, but that minute channels have also been opened up in the outer surface which make it possible for the plastic to impregnate this layer thoroughly. Lower down there is a layer which has not yet suffered much channel erosion and therefore plastic does not penetrate and it leaves no residue on etching. It is much more difficult to explain the thin layer of unetched plastic further down the shell. Its position suggests that it is possibly at the junction of the mammillary and spongy layers and it may be that erosion has taken place up the spaces between the individual mammillary knobs and then along the “clevage plane” between the- iii
a, b, c and d show radial sections giving different etching patterns. A pore cast is also seen on each section and measurements are given. e and f are plastic casts of 1 sq. mm. area of the mammillary layer. g represents 1 sq. mm. area of 1202 partially dissolved away to show the spread of natural stains concentrically from the pores.
- iv Page is blank- 121
two layers. The mammillae themselves have been very much eroded and a brown stain suggestive of erosion runs along the original pieces of shell at the point where the resistant plastic layer of the cleavage plane is left behind by acid.
Another piece of this shell was slowly dissolved away in EDTA (see later) until reduced to about half its original thickness. Fig. 5g clearly shows the result and it will be seen that the stains have developed concentrically from the pore channels. The collar of white material on some of the pores gives the same idea.
Sample 1202 is a very obviously eroded sample and since it gives resistant plastic layers, whereas 1204 is probably not eroded and gives no resistant plastic layers, it may well be argued that in the moa, but not necessarily in other birds, the presence of a resistant plastic layer indicates erosion. If this is so then sample 1205 and 1206 have also suffered erosion, a finding supported by evidence presented earlier. From the entry into an apparently solid shell of large quantities of plastic it would appear that the liquid monomer is able to move along innumerable fine fissures between the calcite crystals; the fissures having been opened up by erosion.
It has already been stated that both appearance and thickness measurements indicate considerable variation in the degree of erosion of the mammillae. These various stages of erosion are shown diagrammatically in Fig. 6 and for comparison a mammilla from an unattacked shell has been drawn. Clearly the protein core is attacked first, but probably not until the membrane has been lost. Erosion then widens the channels in the mammillae and leads to greater and greater breakdown.
FIG. 6.- 122
Diagramatic representation of the successive stages (a-f) of mammillary erosion. a shows an uneroded mammilla with core at the base. f shows the stage where the mammilla has almost gone.
Support for this idea is forthcoming from the plastic moulds of the mammillary surface, Fig. 5. In the uneroded sample 1204 the plastic mammillary cups, each corresponding to an original mammillary knob, were very clear cut with high walls and in each one was a small “hillock” where the plastic had penetrated into the protein core (Fig. 5e). The other samples gave variable results depending on the pieces of shell used, the bigger the hole in the mammillary knobs the bigger the raised piece of plastic required to fill it, but around this raised piece of solid plastic there was always a covering of fibrous white material, indicating that erosion has always gone further than is obvious by inspection. With sample 1202 the mammillary cups tended to be almost filled with a white deposit (Fig. 5f).
It is necessary to stress at the outset of this section that because of the shortage of material it has not been possible to duplicate all analyses, neither has it been possible to use the same piece of shell for two determinations; thus there are bound to be certain discrepancies, for example, the negative value for pore protein in Table 4.
Composition of the Shells.
Tyler and Geake (1953a) devised a method for removing the membrane and cuticle from a shell whilst leaving the protein of the true shell intact. This made it possible to obtain a clearer picture of shell composition than the one obtained when the membrane is left on. They also used an arbitrary method for dividing the true shell protein into two portions, one of which is resistant to prolonged attack by boiling 10% sodium hydroxide solution and the other which is not resistant. They tentatively suggested that the resistant portion is matrix protein and the other is protein in the pores.
The moa shells as received, clearly had neither membrane nor cuticle and may thus represent true shells. On analysis they gave protein values shown in Table 4. They were then treated by the method of Tyler and Geake and values for pore and matrix protein obtained.
Examination of the values for total protein suggests that, if the values were similar in the fresh egg shells then some shells have lost far more than others, and that all of them are considerably lower in value than any other egg shell yet examined. Compare, for example, - 123 the rhea, figures for which are also given in Table 4. The values however, do not appear to correlate well with the degree of erosion because the highly eroded sample 1202 has the same total protein as the much less eroded sample 1205, whilst the probably uneroded sample 1204 has a lower value than sample 1201. There is also a considerable degree of variability amongst the matrix and the pore protein values, but, in the case of the matrix protein the three least eroded shells, namely 1201, 1203 and 1204 give the highest values.
The values for protein are quite different from the organic matter figure given by Liversidge (1880). He found the loss on ignition and then corrected for moisture and carbon dioxide losses, finally arriving at a value of 4.90%. It may be that there were organic compounds other than protein, which the present determination of nitrogen (x 6.25) would not show, but this is unlikely to account for such large differences. The only explanation the writer can offer is that either the fragments analysed by Liversidge had portions of membrane still on them, or that they had been eroded of mineral matter to give a material richer in organic matter. It is, of course, not impossible that the true shell protein for the moa is over 4% and that our samples have lost a great deal, but certainly no egg of any other species examined by us has reached even 3%.
With regard to the mineral matter of the shell the values, on a nitrogen-free basis are given in Table 4. Apart from sample 1202 the calcium values are of the same order as for the rhea and there is very little variation between them. Sample 1202 is somewhat lower, but even this is above 38.0% and is not very different from the values generally found for the domestic hen and many other birds. The carbon dioxide figures are also similar to those for other birds and show little variation between themselves. However, a more detailed study of the two sets of values together shows an interesting point. Tyler (1950) and Tyler and Geake (1953b) showed that there was a highly significant relationship between the weight of calcium and the weight of carbonate in the egg shell of the domestic hen. The equation as given by Tyler and Geake (1953b) took the form:—
Y = 1.118 X -0.0018
where X = weight of calcium and Y=weight of carbon dioxide.
Now this equation indicates that there is an excess of carbonate over that required to form calcium carbonate and this may be associated with magnesium.
Unfortunately, absolute quantities of calcium and carbonate in the whole moa egg shell cannot be calculated, but percentage values give the following highly significant (P <0.01) relationship.
y = 0.963 x +5.17
where x = per cent calcium and y=per cent carbon dioxide for the six samples.
This shows that there is less carbonate than is required to form calcium carbonate and if magnesium also occurred as carbonate the deficiency is even greater.- 124
The magnesium values are lower than for the rhea and, in fact, are the lowest yet recorded for any set of egg shells. If all values are taken there is no relationship between calcium and magnesium percentages, but if sample 1202 is omitted the rest show a significant relationship (just short of P = 0.01), and it takes the form:—
w = 0.16 x -6.12
where x = per cent calcium and w = percent magnesium.
Thus, with one exception the two elements appear to be directly related in quantity, high values for one being associated with high values for the other.
The values for phosphorus are very variable, but the higher ones are greater than any previously recorded figures. There is a highly significant inverse relationship (just short of P = 0.001) between phosphorus and carbonate and this gives the equation:—
z = 10.27 -0.238 y
where y = per cent carbon dioxide and z = per cent phosphorus.
The values for sodium and potassium are so small that they can almost be ignored.
Considering all these relationships together it may be that they are capable of explanation on the basis of erosion effects rather than on the basis of real differences between the original shells. The values suggest that erosion factors have removed calcium and magnesium simultaneously and that carbonate has been removed at the same time, but in proportionately greater quantities if it is assumed that there was an excess of carbonate over calcium to start with. This excessive loss of carbonate however, is counterbalanced to some extent by a gain of phosphate. If these changes have occurred, then the moa shell must have originally had a nitrogen-free mineral composition similar to that of other egg shells, but the six samples have undergone different degrees of change. On this basis sample 1202 has taken up by far the most phosphorus and has lost the largest amounts of other constituents, and it is this shell which is visibly more eroded than the others.
In the last column of Table 4 values are given for the total amount of nitrogen-free material accounted for by calcium, magnesium, carbonate (as CO3) and phosphorus (as PO4). It will be seen that sample 1204 and 1201 which have been judged on many counts to be least eroded give the highest values, whilst 1203 is not far behind. The next two values are for 1205 and 1206 for which there is some evidence of erosion and well below the rest is 1202, which is a badly eroded sample. Of course, the total mineral matter should give 100% and hence there are either other undetermined constituents or else the cumulative errors of determination are negative.
Experiments on Artificial Erosion.
A few experiments were performed in an attempt to obtain some evidence as to how the erosion of the mammillary layer takes place. From descriptions given above the first stage is a tiny hole in the - 125 mammilla which gradually enlarges, but the outer casing of the mammilla always seems to be less eroded so that each mammilla looks like a volcano, but the greater the erosion the lower the walls of the crater.
Pieces of shell from sample 1201 were carefully selected on the basis of negligible erosion, and none of these pieces had holes in the mammillae, which appeared to be unattacked.
A piece of shell was placed in ethylenediamine tetra-acetic acid (EDTA) made up as a 6% solution of the trisodium salt with 6% formaldehyde added. This solution acts as a powerful decalcifying agent and it was thought that there might be some differential etching of the sample. However, at no time did holes appear in the mammillae. The mammillary knobs simply dissolved away until only the stumps were left and the surface looked like a crazy-paving. It is of interest to note that the mammillary surface was dissolved away far more rapidly than the outer surface and this fits in well with the general picture of the naturally eroded pieces.
A second piece of shell was boiled in 2.5% sodium hydroxide solution for 45 minutes and many of the mammillae then had holes in them. This treatment produced no decrease in shell thickness. The piece was then placed in EDTA and gradually the holes were enlarged until at the end of three hours many had reached the “volcano-crater” stage and the piece was similar to the much eroded 1202 sample. This treatment, of course, reduced the thickness of the shell.
A third piece of shell was put into a mixture of equal parts EDTA solution and 10% sodium hydroxide and kept just below boiling point. At the end of three hours each mammilla had a wide but shallow depression in it.
A fourth piece treated with EDTA for two hours and then with 10% sodium hydroxide solution for thirty-five minutes, gave no sign of holes at any stage.
It thus appears that the treatment with sodium hydroxide produces holes in the mammillae and that removal of mineral matter enlarged them but removal of mineral matter first, whether followed by alkaline treatment or not, gives no typical holes.
Ordinary shells may be decalcified completely with EDTA and leave behind a “ghost” of organic material which can be studied histologically. Attempts to do this with moa shells produced nothing but debris in the bottom of the vessel and it has therefore not yet been possible to study the shells from this angle.
It is impossible to form any absolutely definite opinion as to the original form and composition of these moa shells. Furthermore, varying degrees of erosion mask any species differences which may have existed. Nevertheless it is possible to make some tentative suggestions - 126 about these shells and to indicate possible ways in which erosion has taken place.
Measurements of shell thickness, including the respective thickness of the spongy and mammillary layers, coupled with examination under the microscope of the individual pieces so measured, give a very good indication of those pieces which have not been eroded. By this means it was possible to suggest that sample 1204 is the least eroded of all, and that other samples, whilst giving pieces showing no erosion, also gave pieces showing varying degrees of erosion of either one or both surfaces. Sample 1202 was very badly eroded and it is doubtful if samples of this nature can throw much light on the original structure. Replicate thickness measurements on pieces of shell and over many pieces per sample strongly supported the idea that sample 1204 was least eroded, and it is possible therefore to say that the thickness given by this sample, namely 1.40 mm. is a very reliable value. It is of interest, therefore, to note that samples 1201, 1203 and 1205 also gave closely similar values for their thickest uneroded pieces, so that for four samples out of the six the thicknesses are in close agreement. This suggests that the original eggs which gave these four samples may have been of similar size and from similar types of moa.
With regard to the pore structure and distribution it can be said that pores of non-eroded pieces may have single, double or triple mouths and in this respect are like the rhea. On the other hand the depressions in the shell surface into which these pores open may house the mouths of more than one pore. Thus there may be a single pore mouth, two pore mouths or as many as five and with the increase in number of mouths the depressions become more and more elongated. Furthermore, a number of pore mouths in a depression may arise by a variety of combinations of pores. It may therefore be suggested that more extensive measurements than were possible in this work could give valuable information. A statistical study of the proportion of single, double and triple mouthed pores and the proportion of single, twin, triplet and quadruplet mouthed depressions may each be of diagnostic value. However, it must be stressed that careful checking by thickness measurements as well as by other tests should first rule out any question of erosion, before such studies are made. Clearly, if the outer surface is eroded the length of each depression will gradually diminish, and finally all pores will show only one mouth. An example of this has been described above and, taken by itself, this piece of shell could have been quite misleading in this respect.
The mammillary layer suffers erosion to greatly varying degrees, but this can usually be clearly seen under the microscope, and supporting evidence usually comes from the plastic moulds. A small rounded “hillock” at the bottom of each clearly defined deep walled cup is usually indicative of no erosion, but deposits of extraneous material between the mammillae might well modify this picture.
Plastic embedding techniques also show other variations in samples which are indicative of erosion and it would seem that white fibrous plastic material, resistant to the attack of acid is indicative of erosion - 127 even when the visible signs of erosion are not always apparent. Uneroded samples 1201, 1203 and 1204, when embedded in plastic and exposed on a radial face showed a uniform and rapid etching when put in acid, but eroded samples such as 1205 and 1206 showed layers of the white fibrous plastic resistant to acid attack. In sample 1202 there was an additional layer of resistant plastic along the “cleavage plane” between the mammillary and spongy layer. This white plastic material whether in the spongy layer or in the mammillary cups seems to give a very reliable indication of erosion in the moa. It is, however, important to point out that it has not necessarily the same significance in other shells, for many shells give such layers as part of their normal structure.
Chemical analysis indicates different degrees of erosion in the different samples and, in general, supports the idea that sample 1204 is probably the least altered. It may be fairly confidently stated, however, that these moa shells when fresh, did not differ greatly in composition from other egg shells. The biggest difficulty with chemical analysis seems to be that many pieces of the same sample need to be analysed because different degrees of erosion will give different results. Unfortunately this is not always possible.
Arising out of this work, and by bitter experience, it has become clear that every piece of shell which is to be embedded in plastic or used for chemical analysis should first of all be measured for thickness, counted for pores and examined for visible signs of erosion, for it is only from this knowledge that accurate interpretation of the plastic models and the analytical figures are possible and achieve their maximum value.
The failure to produce sections for histological study indicates either that the moa shell does not possess a coherent organic structure or else that this structure is modified or destroyed by erosion.
Finally, there is the interesting problem of the mechanism of erosion. Weathering of samples exposed above the soil could occur but the effects of solvents and micro-organisms on buried samples are probably more important. Considering first the inner surfaces of the samples available it is clear that the protein-rich core at the base of each mammilla is first attacked. This hole is then gradually enlarged and at the same time minute channels are opened up from it, probably along the original lines taken by the threads of protein which radiate from the central core. As these channels are enlarged they will coalesce and the hole itself will get bigger until finally the mammillae are eaten away. This process will give a “volcano-crater” effect at each stage, with the crater gradually becoming shallower. Experiments performed to show the effects of artificial erosion tend to confirm this sequence. The sodium hydroxide and EDTA treatments remove protein and mineral matter respectively, and it was found that holes were only produced in the mammillae when the sodium hydroxide treatment preceded or ran concurrently with EDTA treatment and never when the EDTA treatment was applied first.
At the outer surface there is no easy starting point such as the protein core of the mammillae, and the artificial erosion experiments - 128 indicate that the outer surface was less easily attacked than the inner surface of the shell. However, the radial sections of some plastic embedded samples (1205 and 1206) indicate that erosion seems to proceed along the original threads of the organic matrix, and natural staining seems to follow a similar course. Such shells may show no obvious sign of erosion on this surface but clearly if it becomes extensive then layers may flake off as they have done in sample 1202. The starting point of this kind of erosion could be the surface of the shell, but it is more likely to be inwards from the edges of the broken pieces and also concentrically outwards from the pore walls, as shown again by sample 1202.
The evidence for erosion as shown by the etching of radial sections of shell embedded in plastic must be very carefully weighed. Tyler (1956) showed that the plastic monomer can penetrate many types of shell and give rise to typical bands of plastic when etched. However, none of these shells were eroded. The presence of plastic bands in shells treated in this way is thus not of itself a sign of erosion. It is only when uneroded shells of the moa are compared with eroded shells and when all the other evidence is considered, that it is possible to suggest that in the moa the plastic residue indicates erosion and is not typical of a normal moa shell.
The chemistry of the erosion process is probably related to attack by acidic soil water. Calcium, magnesium and carbonate are dissolved and, in some cases, phosphate also present in the soil solutions, appears to be precipitated. The preliminary removal of protein from the mammillary knobs is probably brought about by microbiological action. Of course, in the soil, there may be other conditions which provide for only microbiological attack or acid attack, but not both. Shells under these conditions may show quite a different picture.
The question of shell membranes has not been discussed because they were completely absent. However, it is possible that they are removed by microbiological action before the attack on the protein cores of the mammillae begins.
The writer is not qualified to judge abrasion produced by man as distinct from erosion produced by other natural agents, but it can be said that there were no obvious file marks on the surfaces of any of the shells studied.
Six samples of moa shell have been examined by methods which included physical measurements, chemical analysis and a study of structure by the use of plastics.
The major feature appears to be that many pieces of shell have undergone different degrees of erosion, but the methods of study used, when taken together, indicate quite clearly which pieces are eroded. This should provide useful information enabling eroded pieces to be avoided when trying to characterise the shells for purposes of comparison.- 129
The maximum thickness of the shells examined is about 1.40 mm., made up of about two-thirds spongy layer and one-third mammillary layer.
The examination of pore casts shows that there may be anything up to three mouths per pore and, further, that the mouths of more than one pore may open into a common depression. It is suggested that statistical studies of these points may be of diagnostic value where larger samples of shell are available.
Chemical analysis indicates that, despite variations possibly caused by different degrees of erosion, the composition of the moa shell is not greatly different from that of other eggs studied.
Finally, the mechanism of erosion is briefly discussed.
I wish to thank my colleague, Dr. D. A. Balch, for carrying out the chemical analysis, Mr. R. Wheeler for the photographs, and, in particular, Dr. H. D. Skinner for supplying the samples of shell.
METHODS OF ANALYSIS.
Pieces of shell were dissolved in 1.ON HCl at room temperature, the solution filtered to remove organic matter, and then made up to volume. Aliquot portions were used for the determination of sodium and potassium with a flame photometer, calcium was determined by titration with the disodium salt of ethylenediamine tetra-acetic acid using murexide as indicator (Schwarzenbach 1955) and magnesium by the method of Hunter (1950) with the modification that after the excess dye had been extracted with the organic solvent a known volume was poured off into a tube containing 5 ml. of absolute ethyl alcohol. Carbonate was determined indirectly by estimating the residual acid after dissolving another portion of shell in 1.ON HCl. A separate nitric-perchloric acid digest of shell was made for the determination of phosphorus by the method of Hanson (1950).
1 Tyler and Geake 1953a; Tyler 1953 and 1956.
2 Tyler 1956.
3 After acid treatment.
4 Includes 0.1 quadruplets, not shown in table.