Prior to the industrial revolution, the main source of primary energy in society was biological - agriculture and forestry, with a significant assist from water mills. The biological energy was used to feed horses (used themselves in ploughing, but also in transportation), as well as agricultural workers. The water-mills were primarily used to mill flour (ie also used in agricultural production, for the most part).
My point is the following - we know that the vast bulk of the pre-industrial population were involved full-time in agricultural production, and so were essentially part of the input energy (they, and their horses, had to be fed in order to go out and till the ground, harvest the crops, etc). Thus, most of the population was essential to the operation of the energy gathering system, and not part of the surplus generated by that system. Peasants by and large had pretty close to the minimum in terms of housing, furnishings, and clothing, and children and the elderly can probably be assumed to be involved in household economies to the extent they were capable. There were basically three places where an energy surplus generated by medieval agriculture could go:
- Population growth
- The church
- Goods and services consumed by the rich, the middle class, and the military
The second factor, the church, we can bound because we know that the church was generally supported by a tithe, in which 10% of food production and craft goods had to be turned over to the church. Now, some of this tithe was used for relief of the poor and hungry, and thus wasn't really surplus. You could also argue that the religious functions of the church were actually necessary for the operation of pre-industrial society, and so weren't surplus on those grounds. Also, I would guess that collection of the tithe was imperfect. But for the purpose of establishing an upper bound, let's treat it that the church was entirely surplus to agricultural production and that the tithe was all collected. Thus we need an EROEI of at most 1/0.9 = 1.11 in order to support the church as a surplus. Let's call it 1.1 as a rough number.
The last factor, goods, housing, etc consumed by the non-peasantry is the hardest to estimate. However, I argue that it must have been roughly bounded by the fraction of the population that lived in towns rather than in small villages and on isolated farms. It's a reasonable assumption that the population of villages was for the most part either directly involved in agricultural labor, or in the production of goods and services immediately required in agriculture (the village blacksmith, the miller, etc).
Now, probably, the population of the towns were somewhat involved in good and services utilized in agriculture also (eg a central function of towns has always been to provide markets to allow trade in goods to occur in the surrounding countryside, and between regions). Still, it's probably a reasonable assumption that non-agricultural production was primarily centered in the town - armorers, furniture makers, masons, etc. And it's probably also a reasonable assumption that the ratio of horses to people was not markedly higher in the town than in the country given that horses were directly used for ploughing and also for transportation of agricultural goods. In any case, only about a third of crop production was fed to horses, so it wouldn't devastate my case if this last assumption was somewhat wrong. Given these assumptions, the energy consumption of town versus country will approximately scale with population - it's unlikely town dwellers ate vastly more than country dwellers - not more than 10-20% more or they would all have been grossly obese (it seems unlikely that town dwellers engaged in more physical labor than country peasants, so surplus food consumption would have gone to fat).
Now, historians have attempted to estimate the fraction of the population living in towns at various times and places. Braudel (p483-484) summarizes the situation as follows ( insertions are my clarifications):
Recent calculations by Marcel Reinhardt conclude that in France in Cantillon's time [the early 1700s], the urban population was only 16% of the total. And, of course, it all depends on the base level adopted. If towns are considered to be settlements of over 400 inhabitants, then 10% of the English population was living in towns in 1500, and 25% in 1700. But if 5000 is taken as the minimum definition, the figure would only be 13% in 1700, 16% in 1750, 25% in 1801. It is therefore evident that all the calculations would have to be repeated using identical criteria, before one could make a valid comparison of the degree of urbanization of the different regions of Europe. At present, all we can do is identify certain particularly low or high levels.(He then goes on to discuss the somewhat higher urbanization in 16th-18th century Holland, but I am going to discount that example because Holland was a small country that was already a major trading hub for all of Europe, and so I don't think can be taken as a reasonable sample).
At the bottom of the scale, the lowest urbanization figures relate to Russia (2.5% in 1630; 3% in 1724; 4% in 1795, 13% in 1897). So the figures of 10% for Germany in 1500 is not insignificant compared to the Russian figures. The same percentage is found in colonial America in 1700, when Boston had 7000 inhabitants, Philadelphia 4000, Newport 2600, Charlestown 1100 and New York 3900. And yet, in 1642, in New York (still known as New Amsterdam) 'modern' Dutch brick was already replacing wood in house-building, a clear sign of growing prosperity. The urban character of these centres where the population was still of modest size is clear to see. In 1690 they represented the degree of urban tension permitted by a total population of 200,000 or so, scattered over a vast area: about 9% of the whole. In about 1750, of the already dense population of Japan (26 million) 22% were already living in towns.
Thus we see figures ranging from about 3% urban population (in Russia) to 20-25% in late pre-industrial England and Japan.
Expressed as EROEI then, and including the 10% church term, we get a range from 1/(1-0.1-0.03) = 1.15 in Russia to 1/(1-0.1-0.25) = 1.54 in England and Japan shortly before the industrial revolution. Since we certainly don't have three significant figures of precision here, let's call it 1.1 to 1.6. Note that, to the extent the church tithe was not surplus, and town economic production was actually required for agriculture, these estimates are probably upper bounds (with some possibility of slightly higher numbers if animal energy usage was higher in towns than I am assuming, or if agricultural peasant households actually had some consumption of goods that could reasonably be considered surplus).
So it's against this backround that one need to consider the introduction of fossil fuels, and their possible replacement now with other forms of energy.
It's also clear that modern biofuel EROEI's are in the same range as pre-industrial agriculture, and therefore are completely unsuited to support an industrial civilization.