OK, I posted my last entry in the Allies & Enemies of Rome forum at R.A.T. (Goths: Moving Whole Populations) and I got a handful of very useful comments and insights.
This is a modified form of my final post over there, and the basis of my next calculations, etc... More comments or help will be welcome!
Yes, I guessed the Huns would just be the warriors and maybe their slaves. If the Alans were also just warriors, then the group of Alatheus and Saphrax would have a higher ratio of warrior-to-civilians... I didn't thought about the selling of slaves and children to the Romans after the crossing of the river, though, good point.
Before crossing the Danube, the Tervingi would have a ratio close to 20% (the deaths of warriors against the Huns would be compensated by the death of the elders and some children due to Winter suffering in the Carpatos).
Let's, for the sake of simplicity, deal with the lowest estimations (some 20,000 Gothic warriors at the Battle of Hadrianopolis, half of whom were Tervingi).
This roughly means some 50,000 Tervingi.
We know many of the elders and weaklings were left behind. They account for some 3% of the population, let's suppose half of them were left behind, so we have some 49,000 Tervingi. Call this number T.
Now, let's try to get an account of what's going on: the roughly 20% of warriors would be among the free men, but let's simplify including all able men, including slaves.
In 'young' populations, age pyramids show that about half the population is under the 15 years old age, so we get as many children as adults.
A familiar core would be comprised, in average, of mother, father, some 4 or 5 sons, one grandparent and a slave or two, although this would vary wildly depending on the economic and social status (the higher the status, the more slaves and the lesser children, in order to avoid heavy fighting for the power).
In average, anyway, we have a familiar core of some 10.5 members. This means that the Tervingi was formed of some 4,666 family cores. Call this number F.
Now, once on the other side of the river, they started to get too hungry and started to sell slaves and sons to keep the rest fed and alive. How many sons per family would have been sold? We can only speculate.
The richer families would have had gold and slaves to avoid selling their younglings, or probably one of the younger girls, while the poorer families would have sold families up to all their sons except the first one (maybe two if the first one was a female, but undoubtly)... It also depends on how much food they got from the sellings, and how long would it have served to feed the rest of the family.
Anyway, we know it was bad enough for people to be at unrest but not as bad as to force them to rebel right away, so I'd say that an average of one to two sons per family would be a realistic number (let's say 1.5).
That means the total population now is close to T - 1.5F ~ 42,000 of which 10,000 are warriors (per the 20% rule before starting to lose civilians).
Now, if the 42,000 Tervingi were able to provide some 10,000 warriors to Hadrianopolis, led by Fritigern, then the rest of the 10,000 were provided by the fleeing Greutungi led by Alatheus and Saphrax, and their Alan and Hun allies.
Let's say that the number of allies wasn't greater than the number of warriors the Greutungi were able to muster, in case of a mutiny. That would put the maximum number of Alans and Huns at 5,000.
Now, in this case, we know that the Greutungi had suffered heavy losses against other Hunnic and Alannic tribes the previous two years. And we know that most if not all of the Greutungi were forced to abandon their lands, so we must expect a big number of population. The Tervingi only decided to leave Athanaric, but we know he kept many retainers on the Carpatos, although we can only especulate. They were enough to allow them to survive against the smaller Hunnic bands, but not so many as to allow them a huge victory. Considering 3,000 would be a big army, I'd say a maximum of 2,000 (about 4 comitives) which would throw some 10,000 more Tervingi, or roughly the 16% of the total Tervingi population (and we aren't taking into account the losses of several skirmishes against the Huns because we are told that Athanaric was pretty good in retreating without heavy losses).
So we have that the total Greutungi population, if of somewhat similar size to the Tervingi, would be some 60,000 people, of which some 12,000 would, originally, been warriors. Heavy losses were to be expected when an army lost some battles, but considering they managed to be fighting for some years, that would mean that either they had the ability to replenish their ranks faster than others, or that they were able to keep their losses relatively low.
Now, the final number should be about 5,000 or bigger, or they would be at the mercy of their allies if they decided to defect them. Let's say the Greutungi suffered heavy losses, about half their warriors.
That would mean the total Greuthungi population would be of 54,000 people, of which 6,000 were warriors. And so we have about 4,000 for the Alans and Huns. Now, before jumping on conclusions, we will take half the civilian population from the Greuthungi, who fell under Hunnic power precisely because they had been let without warriors (simple, I know, but it's a rough estimation).
So, we get 24,000 civilians and 6,000 warriors, about 30,000 Greuthungi.
Summing all this up we have, for the aftermath of the rebellion at Martianopolis, when the Tervingi and the Greuthungi united, but before the battle of Ad Salices in 377:
* 42,000 Tervingi, with 10,000 able warriors.
* 30,000 Greuthungi, with 6,000 able warriors.
Later, about 4,000 Alans and Huns joined in as well. And later on, Athanaric's Goths crossed the Danube and joined the Goths in Illiricum, when they signed the foedus with Theodoric. The proportion of populations are 14 Tervingi to 10 Greuthungi, almost 3 to 2, which would justify the leadership of Tervingi over the whole population later on (first Fritigern, then Athanaric, then who-knows until Alaric is raised reiks and judge in 395 (specially since the kings were elected by the warriors).
All this is, as you can see, highly especulative, however I think it seems reasonable. I have supposed that, as the Goths didn't have siege experience, they weren't able to recover their sons or slaves from their Roman buyers.
So, approaching the Battle of Hadrianopolis, we have a grand totale of 76,000 people, of which 20,000 were soldiers, and about 55,000 were women and children of both sexes.
Now, do these numbers make more sense? I hope that the process I have followed is, if a bit especulative, at least plausible enough, considering the lack of data we have.
The main points are: Tervingi and Greuthungi populations were of similar size before the arrival of the Huns, the Tervingi suffered heavy civilian losses after crossing the Danube, while the Greuthungi suffered heavy military losses before crossing it, and only about half of them managed to escape the Huns.
If we follow this process for the 25,000 highest number of warriors, we would roughly get the following numbers:
* 53,000 Tervingi, with 13,000 warriors
* 36,000 Greuthungi, with 7,000 warriors
* 6,000 Alans and Huns
And a total of 95,000 people, of which 25,000 would be warriors.