All endorheic lakes in arid or semi-arid areas precipitate salts when their salinity overcome a given value. The sequence of precipitation (Eugster and Hardie, 1978) tightly follows the path carbonate-gypsum- sodium sulfate- sodium chloride- magnesium then potassium salts, with important variations linked to the chemical composition of dissolved salts in the incoming waters – chemicals carried on by rare rain precipitations may be considered as negligible in the balance. The occurrence of redissolutions and/or recombination of redissolved ions, diagenesis in older deposits etc…So that practically all lakes present deviation relative to the schematic sequence presented above. Among the evaporitic sequences thoroughly studied, is that presented by the Great Salt Lake in Utah (USA), where mirabilite Na2SO4, 10H2 is a major component of the chemical deposit sequence.
Solubility of mirabilite is 10 g/l at –10°C, 100 g/l at O°C,and 490 at 25°, 730 at 33°C, and, following some authors, 930 at 34°. Above 35°C, it looses its important hydration sphere and transforms into thenardite, Na2SO4, the anhydrous form of sodium sulphate, which is powdery and easily blown away, and which constant inhalation- as for populations around the Aral Sea, submitted to such an aggression, is the cause of many respiratory diseases. At low temperature (under 0°C), halite (NaCl) is much more soluble than mirabilite. Mirabilite begins to form at 100 g/l salinity in winter, with coprecipitation of some halite  (NaCl). It redissolves in summer and is eventually transformed on shores in powdered thenardite, which is easily blown away by wind from the shores, as water level is lower. This occurs frequently today on the drying swampy shores of Aral.
Fossil mirabilite sequences in sediments of the Aral sea had been studied in detail  by Rubanov (1977,1987) . He discovered them in the deepest parts of the Sea (fig 1). These beds, are covered by a cap of gypsum, then silt for about one meter, deposited since the return of Amu Darya and Syr Darya waters to the lake , a few centuries ago, after the lake regression which is now well attested not only by historians (Kes, 1990) but by the recent discovery of tree stumps and buildings  around 15-20 m under the 1960 water level (Boroffka, 2005) . Rubanov predicted (1987) that in the future some bays of Aral mirabilite should form as it did in former regressions.
Since 1960, level of water decreased by about 24 m, with the total salinity of South Aral water shifting from about 10g/l in 1960 to more than 90 g/l in 2005, having lost most of its calcium as calcite and gypsum (see Letolle and al., 2004).
Late Tchebas bay (fig.1 and 2), – where Rubanov found one of those past mirabilite deposits was up to june 2005 linked to the Bolchoie More by a narrow straits less than 0.5 meter deep, which and became a separated from the main Water body in august 2005 (fig 2), becoming a lake about 15 km on 7 km, with a depth of less than six meters, now fed only now by rain and small temporary rivulets . The geometry of the past bay , just before it became an isolated lake, is now the same as it was during the previous historical regression, when a small inlet provided it with just enough water for evaporation to lead to the stage of mirabilite precipitation during winters at the bottom, when water temperature could be as low as O°C without freezing.
In fact, due to the rebuilding in 2005 of the dam separating the Small Sea from the Big Sea, the latter level goes on lowering , so that Tchebas bay represented the past configuration of an evaporating pan for a few months only.
Fig.1 Position of evaporite beds : 1- Gypsum thicker than 0.5 m; 2 - Mirabilite; 3- Past shore before 1960.
Channels are the present hydrographic net of the present lake. Kerden archaelogical site is indicated by "x".
From Rubanov’s estimations, the fosssil mirabilite bed is elliptic with 15*7.5km axes, that is about 90 km2; for the known thickness of beds (at last 1m), the volume of M would be 45 Mm3;, thence 85 Mt (d=1.89), coming from evaporation of 25 km3 “normal Aral Sea water”
Fig.2 : Tchebas bay as seen two months apart (from Spotimage and Modis archives)
Accumulation of mirabilite imply several conditions : a regular and not-too-high input of external water with the possibility of evacuation of more soluble brines, low temperatures in winter, and specific conditions for mirabilite not to redissolve in summer. This last condition implies that the brine precipitating mirabilite infiltrates in the sediment, which, in any case will not attain in full summer a temperature of 35°(destruction of mirabilite). Only a part of potential mirabilite can be kept, due eventually to infiltration in bottom underground aquifers redissolving and carrying away Na+ and SO4= ions. Therefore, the mirabilite beds discovered by Rubanov cannot be " primary" “ bottom “ deposits, as temperature of water in shallow waters during summer evaporation may well be above that of mirabilite stability, giving thenardite blown away to the south west as it is today.)
Fig3: Sedimentary cross section mirabilite in pink and gypsum in orange adaptated from Rubanov , Lit. Poleznie isko., 1,1984,117-125)
A theoretical evaluation of volume of salts deposited from Aral water with a starting total salinity of 83.7 g/l and the composition measured by Friedrich and Oberhansli (2003) in Tchebas bay in August 2002, when the basin was yet opened to the Bolchoie More, corresponding to the chemical state of the Aral Sea water at this regression stage : their chemical data is a good starting reference for the water which entered Tchebas when it was in the several centuries ago an evaporitic pan. This water has already lost most of its calcium as carbonate and sulfate.
The use of molar volume and density of various deposited minerals, calcite, gypsum, then mirabilite etc., gives for 1 m evaporated the following value for deposited “ potential “ minerals sequence :
|Anhydrous mineral||Density g/l||Vol in m3 of salt coming from one m3|
of Aral sea water at 83.7.g/l salt solution
|Thickness of salt deposit in mm|
|CaCO3||2.71 (as calcite)||0.2|
|CaSO4||2.32 (as gypsum)||0.002210 (as gypsum)||2.2 (as gypsum)|
|Na2SO4||1.89 (as mirabilite)||0.006270||6.3 (as mirabilite)|
|NaCl||2.5 (as halite)||0.009700||9.7|
|MgSO4||1.68 (as epsomite)||0.135000||13.5|
|MgCl2||1.56 (as bischofite)||0.005670|
|K salts (for memory)||About 0.000200|
As salts concentrate, the solution percolate through the bottom. In shallow basins, as was Tchebas at that time, solubility of mirabilite decreases considerably in winter , whereas the solubility of halite and other potential salts remains high. Just a little halite coprecipitate with mirabilite. After winter, temperature rises at bottom, and part of mirabilite redissolve and establish a system solid-solution in the sediment . Temperature at the bottom certainly will not reach the temperature of dissociation of mirabilite to thenardite (36°), so that newly formed mirabilite from the input water to the evaporitic basin , infiltrating, will add to the remnant of deposited mirabilite from the precedent year. This is the process described by Eugster and Hardie(1978, p.245.):“most of the of this mirabilite is redissolved during the spring, but considerable accumulations have been found in the subsurface sediments, (Herdley, 1962)”
One year summer evaporation of 1m deep laguna brine (a maximum, as evaporation flux diminishes with the shift in salinity) evaporation shifts enough concentration to precipitate mirabilite in winter. In spring, mirabilite dissolves again and infiltrates further down in the sediments, as other dissolved salts, and stabilize there again during the next winter. In depth, seasonal temperature variations should be much attenuated and the speed of mirabilite remobilization at depth also.
Surimposited to this mechanism, one must consider, in order for the basin not to be filled rapidly with incoming salts from the feeding inlet, that saline water percolate down to deep aquifers, but the deposition of salts in the open porosity of sediment could slow the descent and accelerate the formation of thicker deeper solid deposits. In fact, the four basins of the Aral area where mirabilite formed are located on the big North-South fault system which determinates the western Chink, Vozrojdenie "horst", Kulandy peninsula etc. , which have been considered by all hydrogeologists as the main underground way of escape of salts, down to the depth of 3500-4000m: this mechanism was (and is always) is supposed to regulate the chemical composition of the whole Aral Sea when the lake was in steady state (deep aquifers waters are always enriched in Na, K and Cl with regard to surface running waters)
How much time was needed to form Tchebas mirabilite bed?
The table above gives a theoretical thickness of 6.3 mm of solid mirabilite from evaporation of 1m of 84 g/l water - consider this value as an order of magnitude. We do not know the precise thickness of the mirabilite bed,-at last 70 cm following Rubanov, who could not get a longer core- before re-invasion by river waters, so that it would need at last 10 or 15 years to form the known thickness of the bed. But due to the fact that much of the precipitated mirabilite disappeared downwards through the mechanism depicted above, only a fraction of total deposited mirabilite has been kept today . The indisputable conclusion is that deposition of mirabilite in the Tchebas “evaporating basin”asked for a long time during which , Bolchoie More stayed for many years at a level which was approximately the same as in june 2005. This conclusion is unavoidable, and explains that at that time shore populations had the opportunity – and time- to build in the XV-XVII th centuries villages, cult buildings (Kerderi "mazar") of an elaborate style on the past bottom, dried as it is today. Input from the rivers was the order of magnitude of which was approximately the same as today’s one. Gypsum was the essential precipitate product everywhere in South Aral, as it does today. In fact, at that time, the geography of Southern Aral was the same as today's, with Tchebas acting as an " accumulation kidney", in the sense of zoologists, to eliminate excess salts: this would explain why mirabilite is not found at that time in the south eastern basin, but only gypsum. Note that the regression at this time corresponds with the so-called Paskevich drowned terrace.
Presently , the South Western is basin is fed essentially by a short northern channel (fig.4) about 20 km long and 3 km wide,less than 1m deep, except a small deep of 6 m in the eastern part of the channel (Zavialov, pers. com.; currentometry in the channel is presently in course) .The only notable water source, apart of the present northern channel, is diffuse drainage water from the Amu Darya delta in the south. Past conditions for the deposition of mirabilite imply that the water input was smaller as the present one (2005), and therefore that the northern channel was itself almost or completely closed: therefore it seems that in order for mirabilite to deposit, the general level of Bolchoie More should have lowered one or two meters to lessen much more the flow to and fro the western through and the Eastern basin, before the separation became complete. When the western trough became autonomous, and was fed essentially by a residual western channel of Amu Darya following the Chink, the reality of which has been recently established by Yagodin (2002): this western basin evaporated further as Tchebas Bay did some time earlier.
fig.4: False colors view of Kulandy-Vozro channel (summer 2005); in blue, depth of water less than 0.5 m.
Age of the “western mirabilite” is therefore older than the Tchebas mirabilite, and it is hoped that new radiocarbon data on the west basin allow to specify its age.
As concerns the South-Eastern Basin, (Rubanov, 1982, Maeva, 1982, Maev et al., 1991) mirabilite is not found, but peat, a first bed followed by a second one in the southern part, covered by gypsum. The likely reason for this beds may be that enough water, either surface or unerground waters, arrived from north ans south-east to stabilize salinity to a value under 100g/l; at a level of about –25m under 1960 watermark. The southern peat bed was C14-dated by Rubanov at about 450-500 AD ; which could indicate the strong regression attributed to the Hunnic invasion was more important that the later one (approximately XVIth-XVIIth centuries AD, see Boroffka et al.), related to the Kerderi site, which could then be contemporary with the upper peat bed, shown on a figure in Maev et Maeva, younger than the southern one published earlier by Butakov in 1982. Redating these peat beds should be very useful.
The small northern basin of Aral Small Sea contains also two mirabilite formations separated by a levee, also discovered by Rubanov. During a former regression (not precisely, or even approximately, dated), this basin was separated from the southern basins as soon as the level of water fell under 13 m under 1960 level (the depth of past Berg straits between Big Sea and Little Sea of Aral) , and had a distinct behaviour. Syr Darya delta branches have often changed in the past, naturally and artificially), so that Maloye More basins were sporadically fed by a northern branch of the Syr Darya delta (reanimated in the 1980’s), or perhaps only by water coming from the Bolchoie More through Berg straits. Chronology of past channels of Syr Darya is unknown before 1830.
Western Maloye basin (Chechenko Bay) precipited mirabilite up to the time it was itself separated from the eastern basin (comprising Butakov Bay) when the water level went under 17 m; , then the eastern Maloye basin only precipitated mirabilite. Up to now lack of radiocarbon measurements forbid to give the chronology of these events.
The discovery of Kerderi archaelogical site is essential as it indicates that at this time there was a (small) flow running from a southern branch of the SD delta, and, together with south-eastern channels of the same river, fed the south eastern basin after Tchebas was semi-isolated.
A possible reconstitution of events of the last historical regression could be :
-A first step of Aral dessication, most of the water input ceasing abruptly as since 1960, except 1- a small quantity going to the Bolchoie More from Amu Darya and /or the southern past channel of Syr Darya, called Yani Darya (used to day to convey drainage water from the Syr Darya valley) , and 2- also a small quantity from Syr Darya to the Small Sea, which became shut off to the Big Sea rapidly; the basin of Small sea receiving too little water went dry and gave mirabilite, successsively in Butakoff Bay and Perowsky Bay, as imaginated by Rubanov twenty years ago (Rubanov 1987, conclusion of his book); these evolution being independent of the evolution of the Big Sea.
- Then the Big Sea level went down to –22 –23 m , with episodic input to the separated Tchebas , and a more or less long stay about this altitude during a long time, perhaps one century or more being necessary to build the Tchebas mirabilite. Each spring, as the level of " remaining" big Aral shifted a few decimeters for water to invade Tchebas bay and precipitate mirabilite.
- Then the level went down again by a few meters, Tchebas bay was completely closed, the Vozro-Kulandy straits shut off, and just enough water went to the eastern basin to feed the small “Rubanov oasis” in the eastern south basin, which went to (almost) dryness, at about –26 m. A small quantity of water yet arrived from the south-west Urgench branch of Amu Darya to the western trough, possibly leading to precipitation of the biggest mirabilite deposit of the Aral Sea. Here, the small valleys on the west slope of Vozrojdenie " ridge" could collect mirabilite as soon as it formed when the water level went down to 0 m asl, and the Tchink cliffs on the west forbade the elimination of mirabilite by north- eastern dominant winds . If Aral is to have the same evolution, in the years to come, the same scenario should occur.
In any case, the building of the past mirabilite deposit in Tchebas bay, taking into account the losses by infiltration and /or sweeping away by aeial erosion , imply the evaporating system worked there for many years, perhaps more than one century, with the water level of South Aral remaining more or less stable around its present level (30m asl) for many years. Such conditions could not permit deposition at the same time of mirabilite in the western basin as the Kulandy -Vozrojdenie straits provided a permanent – though certainly small - input of water from the eastern basin, except if this flow was small enough in comparison with the total evaporation flux on the western basin. When the level of eastern basin went down a little more (order of 1 m), feeding western basin will be more or less intermittent, with seasonal variations - South - Eastern basin the situation will resemble what it was a few centuries ago, if the present water discharge to Aral is not changed: it may be predicted that in a few years, Western basin will again precipitate mirabilite.
The chronology of deposits of big salt lakes already studied in the world (Utah Great Salt Lake, Dead Sea, Chott El Djerid etc..) is in fact rather mostly unknown, except, as in the case of Aral, through some preliminary studies of past terraces. It would be eminently desirable that long corings- ten metres or more – should be made in a few selected places of Northern Aral –as this was done a few years ago north west of Muynak in Amu Darya past delta – to complement the past campaings of V.I. Rubanov.
 NaCl (solubility 355 g/l at 25°C, a little less at 0°C) begins to precipitate at the same time as mirabilite, up to the final deposits of epsomite and astrakhanite, when considering only the species known to form along the dessicated shores of eastern Aral, where total salinity sometimes overtake 350g/l (Rubanov, 1994) Its; We have not to consider here lower temperatures.
 Mirabilite is commonly found in drying up solonchaks of Central Asia , where dissolved gypsum and halite (CaSO4, 2H20, and Na Cl), are very common  (see Rubanov I.V. 1978), and on playas, especially on the dry shores of the Aral Sea. (Rubanov, 1994). Rubanov, during its numerous cruises on the Aral Sea, gave lithological cross-sections (fig 2); this was the first objective proof , with the dicovery of drowned terraces, that at some time in the past, occurred a big regression of Aral, in harsh conditions of dryness, cf Brodskaya N.G., 1952.
 All dated from the 16-18th centuries AD by C14 measurements.
 which was studied by Lepechkov and Bodaleva as early as 1952.
 In Great Salt Lake, precipitation of mirabilite begins at 4-6°C. A mean value of its chemical composition, (taken from Encyclopedia of geomorphology, Fairbridge R., ed, Reihold publ, 1968) is: SiO2: 248; Ca 241; Mg 7200; Na 83600; K 4070; HCO3 241; SO4: 16400; Cl: 140000; S=254000 (in ppm). Aral Sea is yet far from this total salinity and its chemistry is quite different; but UGSL precipitates mirabilite every winter , and a good part is preserved in sediment. One may think that in its final state, Aral will stabilize around 300-350g/l. if things go on the present way,
 We do not consider here the role of redissolution by Aral water when it re-invaded slowly Tchebas at the end of the regression, carrying water which deposited gypsum and silt, creating the " waterproof" cap protecting mirabilite from remobilization to free water upwards.