Contents
- Yield information contained in medieval manorial accounts
- Yield information assembled in the Medieval Crop Yields Database
- Accuracy and representativeness of the Medieval Crop Yields Database
- Scope and coverage of the Medieval Crop Yields Database
- How the annual chronologies have been reconstructed from the raw data
- Outstanding priorities and future plans
- Selective bibliography (in date order of publication)
Yield information contained in medieval manorial accounts
The many thousands of surviving medieval manorial accounts (sometimes known as compotus rolls and in their enrolled form as Pipe Rolls) contain all the information necessary for the precise calculation of the yields of specified crops, on named demesne farms, in dated years. Each account enumerates the cash and stock received and expended on a single demesne farm managed by or on behalf of a manorial lord over the course of an agricultural year, usually from Michaelmas (29 September) to Michaelmas. Typically, each account records the amount of grain (both threshed and as yet un-threshed) received from the previous year’s harvest and the quantity of seed sown in preparation for the next harvest (see ‘Woodhay 1254-5 grange account’). The information is hand-written on parchment in abbreviated Latin using Roman numerals and the form of the entries is usually formulaic so that with a little practice they are not difficult to interpret. The following extract recording the amounts of barley (Ordeum) received and expended in 1378-9 on the Battle Abbey manor of Alciston in East Sussex (East Sussex Record Office, SAS/G44/34) is an example of one of the more enigmatic types of entry that can be encountered.
It may be transcribed in its abbreviated form thus:
| Ordeum Resp’minus iiijto xv qar vij bz | Et de CCCxxv qar iiij bz ordei rec’ de ex’ gangie per minorem bussell’ titur’ & vent’ ad xxij bussell’ ut sa ac ead’m mensa. Et de xiij qar v bz pa r’ de c’lis eorund’m. Et de xv qar iiij bz r’ de vas’ titur’. Et de v qar j bz pc r’ de c’lis eorund’m. Et de lxix qar ij bz r’ de eod’m ex’ tritur’ & vent’ per opera cust’. Et de iij qra iiij bus’ pc r’ de eod’m ex’ tritur’ per fam’los.
Sma CCCCxxxij qar iiij bz iij pc De quibus in se’ie sr’ Cxxviij acr’ t’re d’nice: un’ in campo in parte Orientali Curie lx acr’; in campo in parte Boriali Bosci xxvj acr’; apud Telton’ in div’c’ parcell’ xx acr’; in parvis parcell’ in Alciston’ xxij acr’ — Cxvj qar iiij bz sr’ acr’ vij bus’ pc plus in toto iiij bz. In cons’ se’iatoris j bz’. In lib’ fam’lorum ut inferius ut in lxxj qar iiij bz di’ maioris mens’ iiijxxj qar vj bz minoris mens’. In vendic’ ad ganar’ de Bello ut patet per j tall’ xxxiij qar. In vas’ tiiur’ cu’ c’lis eorund’m xx qar v bz pc. In Bras’ fus’ ut inferius Clxxvj qar iiij bz. In vendic’ pro exp’n’ canu’ d’ni ut patet per indentur’ ut in vij bz di’ maioris mens’ j qar di’ bus minoris mens’. In cust’ porcorum, porcell’, auc’ campestr’, & al’ per totu’ ann’ iij qar. Sma qz supa Et eqz |
Rendered into extended Latin as follows:
Ordeum | Et de CCCxxv quarteriis iiijbussellis ordei receptis de exitu grangie per minorem bussellum trituratis et ventilatis ad xxij bussellum ut supraac eadem mensura. Et de xiijquarteriis v bussellis peca receptis de cumulis eorundem. Et de xv quarteriis iiij bussellis receptisde vasis trituratorum. Et de vquarteriis j bussello peca receptisde cumulis eorundem. Et de lxix quarteriis ij bussellis receptis de eodemexitu trituratis et ventilatis per opera custumariorum. Et de iij quarteriis iiij bussellis pecareceptis de eodem exitu trituratis per famulos. SummaCCCCxxxij quarteria iiij busselli iij pece De quibus in semine super Cxxviij acras terre dominice: unde incampo in parte Orientali Curie lx acre; in campo in parte BorialiBosci xxvj acre; apud Telton’ in diversis parcellis xx acre; inparvis parcellis in Alciston’ xxij acre — Cxvj quarteria iiij busselli super acram vij bussellipeca plus in toto iiijbusselli. In consuetudine seminatoris j bussellus. In liberationefamulorum ut inferius ut in lxxj quarteriis iiij bussellis dimidio [bussello] maioris mensure iiijxxj quarteriis vjbussellis minoris mensure. In venditione ad grangerium de Bello ut patet per j talliam xxxiij quarteria Invasis trituratorum cum cumulis eorundem xx quarteria v busselli peca. In Brasio fuso ut inferius Clxxvj quarteria iiij busselli. Invenditione pro expensis canum domini ut patet per indenturam ut in vijbussellis dimidio [bussello] maioris mensure j quarterio dimidio bussellominoris mensure. In custibus porcorum, porcellorum, aucarum campestrium etaliorum per totum annum iij quarteria. Summa que[recte: ut] supra Et eque |
And translated thus:
Barley | And for 325quarters 4 bushels of barley received of the issue of the grange by the lesser bushel threshed and winnowed at the 22nd bushel as above and by the same measure [i.e. for wheat]. And for 13 quarters 5¼ bushels received from the heaps of the same. And for 15 quarters 4 bushels received from the containers of the threshers. And for 5 quarters1¼ bushels received from the heaps of the same. And for 69 quarters 2 bushels received from the same issue threshed and winnowed by the works of the customary tenants. And for 3 quarters 4¼ bushels received from the same issue threshed by the famuli [i.e.full-time farm servants]. Total: 432 quarters 4 bushels 3 pecks Of which, in seed sown upon 128 acres of demesne land, of which in the field to the east of the manor 60 acres; in the field to the north of the wood 26 acres; at Tilton in various parcels 20 acres; in small parcels in Alciston 22 acres —116 quarters 4 bushels, per acre 7¼ bushels plus 4 bushels in all. In the custom of the sower 1 bushel. In payment to the famuli, as shown below, for 71 quarters 4½ bushels by the greater measure, 81 quarters 6bushels by the lesser measure. Sold[i.e. paid] to the granarer at Battle as shown by one tally, 33 quarters. In the containers of the threshers with their heaps 20 quarters 5¼ bushels. Made into malt as shown below, 176 quarters 4 bushels. In sales for the expenses of [i.e. expended on] the lord’s hounds as shown in an indenture, namely for7½ bushels by the greater measure, 1 quarter ½ bushel by the lesser measure.In maintaining the pigs, piglets, field-geese, and others for the whole year3 quarters. Total: as above And quit |
Note that:
| [A] | account is given separately of barley threshed and winnowed by (i) hired piece-workers paid 1 bushel for every 22 bushels threshed, (ii) customary workers, and (iii) full-time farm servants or famuli. |
| [B] | the payment in kind to the piece-workers of 1 bushel in every 22 threshed and winnowed appears first are receipt and then as an expenditure in the charge and discharge sections of the account. |
| [C] | when initially threshed and stored in the granary the grain was measured with a heaped bushel but then, when it left the granary, with a struck or razed bushel (i.e. one in which the grain has been struck level with the rim of the bushel measure); by convention, the difference between these two measures is referred to as the ‘increment of the granary’. These are the ‘greater’and ‘lesser’ measures referred to in the account, the latter yielding approximately 7.5 per cent ‘more’ grain than the former. |
After threshing and winnowing, the amount received at Alciston from the harvest of 1378 was 432·59 quarters; of this, 116.5 quarters was used as seed the next year. It follows that two consecutive accounts — the first recording the amount of grain sown, the next, the amount harvested — are usually required for the calculation of yields. In this case the 1377-8 account records that 115 quarters 3 bushels of barley had been sown, giving a yield per seed of 3.75 for the barley harvest of 1378. From the late thirteenth century, however, the medieval auditors often made their own calculations of yields and noted these in the margin of the account. With this information it becomes possible to reconstruct the ratio of the grain harvested to the grain sown from information contained in a single account.
Thus, Battle Abbey’s auditor recorded that the amount of barley harvested in 1378-9 was 15 quarters 7bushels short of 4 times the amount of seed shown in 1377-8. Since the total quantity of barley harvested was 432 quarters 4¾ bushels, the auditor’s calculation implies that [((quantity harvested) + (15 quarters 7 bushels)) ÷ 4] had been seeded the previous year, i.e. 112 quarters 1 bushel (in fact, it was 115 quarters 3 bushels). The auditor’s yield therefore implies a yield per seed of 3.86. In this case, the difference between the auditor’s yield and the calculated yield is probably a product of the idiosyncrasies of measuring grain at Alciston with a mixture of heaped and struck bushels.
When the manors concerned belonged to perpetual institutions, notably conventual, collegiate, and episcopal estates, it is sometimes possible to track yields on the same demesne over periods spanning decades and sometime seven centuries. Here is a reconstruction of the annual chronology of barley yields at Alciston between 1335 and 1492based upon the evidence of consecutive accounts, when available, and auditors’ yields when not.
Yields on this demesne can be measured over such a long period because the manor belonged to a perpetual institution — Battle Abbey. One of the great strengths of the medieval yield data is that on the best documented manors it is possible to track yields over periods spanning decades and, in the case of the estate of the bishops of Winchester, centuries (this is a longer time-span even than that covered by the longest running results from the Rothamsted agricultural research station, the oldest such research station in the world, where experimental planting began in1843). The same sources can also be used to reconstruct the scale and composition of the cropped area, the rates at which seed was sown, the relative numbers and types of livestock kept, and, less easily, the size and nature of the labour force and grain output per worker.
Nevertheless, as the example from Alciston given above illustrates, for a variety of reasons working with this information is rarely completely straightforward. The clearest advice on the range of things to be alert to in the calculation of yields is given by J. Z. Titow, Winchester yields: a study in medieval agricultural productivity (Cambridge, 1972). It is repeated here verbatim:
The following is a schematic presentation of in-coming entries which may been countered under each crop:
| 1. | Grain remaining from preceding year. |
| 2. | Issue of the grange (exitus grangie), with, or without, the increment of the granary. The increment was the difference between produce measured in heaped bushels as it entered the granary after threshing, and the total quantity of grain dispensed from the granary measured in struck bushels (or the difference between full heaped and less-fully heaped bushels). Sometimes, as in early accounts, the issue of the grange was already entered with increment included, or, as in later accounts, it was calculated in struck bushels as well and there was no increment accordingly. |
| 3. | Grain received from elsewhere. |
| 4. | Grain received from the peasants in various customary dues. |
| 5. | Grain bought. |
| 6. | An estimate of grain given to manorial servants in sheaves. |
| 7. | An estimate of grain given to manorial servants as sown acres. |
| 8. | An estimate of grain given to animals in sheaves (usually applicable to oats only). |
| 9. | Poor-quality grain separated mechanically from the better quality grain. This is most commonly met with in the case of wheat and the inferior grain (curallum) is usually entered in a paragraph on its own, following the main entry: thus, although it may not appear under the main entry at all, it must nevertheless be added to it when calculating total produce. |
| 10. | An estimate of grain, if any, remaining unthreshed into the next accounting year.In the earlier accounts this is usually given after the out-going entries and must be added to other produce when calculating total produce. In later accounts it is usually given both on the in-coming and the out-going side and care must be taken not to count it twice when calculating total produce. |
| 11. | Oneratio, a quantity of grain in which the accounting officer was amerced to bring the total to an expected figure. This type of entry is nearly non-existent on the Winchester estates before 1350 but almost constant thereafter when it can be very substantial indeed. |
It will be seen from the above outline that to calculate total produce in any given year it is not enough to take the exitus only, (though it may frequently be the largest or the only item on the in-coming side); other entries which are in the nature of produce must be added to it and those which are not, ignored. Thus items (2), (6), (7), (8) and (9) must be added together to obtain total produce and particular care must be taken not to overlook curallum and any corn left unthreshed if not already included on the in-coming side.
Three other pointsmust be made in this context. Total produce calculated from manorial accounts is total produce less tithe which was normally collected in the fields; on one or two occasions when it was deducted after the grain had reached the granary I have subtracted it from the total for consistency’s sake. Secondly, in the first half of the thirteenth century it was a not uncommon practice on the Winchester estates to sell the corn, partly or wholly, in grosso, that is, before it was threshed. When this happened, no calculations of total produce are possible and such years have to be disregarded, but the grange accounts sometimes forget to mention partial sales in grosso thus giving the impression that the account records the whole of the produce; checking against the sales of grain in the income part of the account brings such omissions to light and should be carried out as a precautionary measure. Finally, a practical problem arises in connection with estimated quantities of grain remaining unthreshed into the next year. Is one to count them at the estimated value, i.e. as given in the current account, or at the exact value as given in the following account? The quantities entered in the following account are usually somewhat lower than the previous year’s estimate; this is probably due to the more exact nature of the measurement, but it could also be due (as it is on rare occasions explicitly stated to be) to loss and deterioration of grain which had lain in stacks for a long time. Since the former possibility seems more likely than the latter I have counted unthreshed corn at the measured value as given in the following account; when such accounts are missing one is forced to use the estimated value despite its explicit inexactness.
On the out-going side of the grange account only two items are of interest to us in this context: the quantity of grain sown and the number of acres over which it was sown. It is over the latter that we again encounter a serious problem. The Winchester accounts, though they do not always make it explicit, use two different acres: the measured acre (acre mensurata per perticam, or, more usually, acra per perticam) and the customary acre (acra sicut jacet). Up to 1232, and occasionally after that date, acres in the grange accounts are unqualified but it appears on examination that they must have been in fact customary acres on practically all the manors. In 1232 all manors of the bishopric went over to recording their acreages in the grange account in measured acres, but a number of them reverted to the old practice subsequently and continued in it until 1320. On most Winchester manors the customary acre was roughly half the measured acre but on a few manors the disproportion was much greater. Since, however, it is impossible to convert customary acres into their measured equivalents with any degree of exactitude, calculation of yields per acre should be, in my view, restricted to years in which measured acres are used. Observance of this restriction limits considerably the number of years for which calculation of yields per acre can be made and this is the main reason why this study is made primarily in terms of yields per seed. An additional reason is that the medieval administrators of the Winchester estates themselves seem to have thought in terms of yields per seed rather than per acre; whenever they made calculations of yields on the margins of grange accounts, as they did frequently in the late thirteenth and early fourteenth centuries, these were invariably multiples of seed.
(Titow, Winchester yields, 7-9)
The following additional points made by Titow should also be noted:
In calculating the total produce everything which is in the nature of manorial produce has been included. This includes such items as curallum, increment of the granary, and corn given away in sheaves to animals, reapers, and manorial servants. Since some of these items began to be recorded only later on in the period, the total produce in the early part of the thirteenth century is likely to be under-recorded.
Quantities of corn remaining unthreshed at the end of the accounting year have been added to the total produce, whenever applicable, at their threshed value, as given in the following account roll; when there are no consecutive account rolls extant, they have been added at their estimated value.
Whenever necessary, ‘new grain’ (novum granum) has been subtracted when calculating the total produce. The term novum granum is at first used synonymously with ‘this year’s grain’ as opposed to ‘old grain’ (vetus granum) which is used synonymously with ‘grain remaining from the previous year’ (granum de anno preterito). Towards the end of our period (i.e. the mid fourteenth century), however, it acquired a new meaning and was used to denote grain ‘poached’ from the new harvest gathered in just as the accounting year was closing. Such quantities have to be excluded from the total produce and added to the total produce of the following account roll where they are usually not recorded.
Oneratio super compotum, when relating to productivity, has been excluded throughout when calculating the total produce. It is, however, very rare before 1350 (but common thereafter).
No allowance has been made for tithes throughout. On the three exceptional occasions when tithes were paid out from the granary instead of being deducted in the fields, they have been subtracted from the total produce before yields were calculated.
(Titow, Winchester yields, 36-7)
On the key issue of dating (i.e. harvest years versus accounting years) Titow has the following to say:
Since the Account Rolls run from Michaelmas to Michaelmas each year of account includes parts of two calendar years; this creates some confusion whenever dates are mentioned. The practice adopted here is to refer to each period covered by an annual account roll by the date of its closing Michaelmas. Thus, for example, to say that some event took place in the year 1245 means that it is recorded in the account roll running from Michaelmas 1244 to Michaelmas 1245. In so far as yields are concerned this means that all years quoted are harvest years, i.e. the yield of 1245 is the yield of the harvest gathered in the year 1245. It is worth noting that this is contrary to the practice, adopted by Lord Beveridge in his articles on the Winchester Rolls, of dating events by the opening Michaelmas of the account roll; this means that information about an event said by him to have taken place in the year, for example, 1245 comes from the account roll running from Michaelmas 1245 to Michaelmas 1246. Lord Beveridge says that his years are harvest years but this is both erroneous and confusing. For example, the account roll for the fourth year of Peter des Roches which ran from Michaelmas 1208 to Michaelmas 1209 is referred to as the harvest year of 1208. This is incorrect; it is true that this is the account roll in which the produce of the harvest of the calendar year 1208 is recorded but the actual harvest itself fell within the period covered by the preceding account roll and all the operations, expenditures, etc., connected with it are recorded in the preceding account roll, i.e. that running from Michaelmas 1207 to Michaelmas 1208.
(Titow, Winchester yields, 35)
This database retains Titow’s recommended method of dating yields by harvest year and Professor D. L. Farmer’s yield calculations for the Winchester estate (which he dated by accounting year) have been re-dated accordingly.
Yield information assembled in the Medieval Crop Yields Database
Even for the most experienced historian, extracting yield information from medieval manorial accounts is painstaking and slow. The task of transcription can be expedited using purpose-designed data-entry forms (see ‘Example of a data-inputting form’) and the calculation of yields from the quantities of grain seeded and harvested can be mechanised by entering the relevant information into a spreadsheet (although care needs to be taken to ensure that this does not generate spurious results in the light of all the gaps and ambiguities that the original data inevitably present). With these aids, a full and productive day in the archives might yield data from two-dozen accounts amounting to perhaps 75-100 individual yield observations but if the originals are worn, damaged, and difficult to read that return might be halved. Moreover, most microfilm and microfiche copies are slower and less reliable to work with than the manuscript originals. There are therefore real logistical limits to what a single researcher working unaided can achieve.
The database of approximately 30, 000 individual grain yield observations assembled here is therefore the product of several historians’ labours undertaken over a considerable period of time and represents the equivalent of at least 2 years of intensive archival work and associated data inputting. The sources of these yield calculations — over two-thirds of which have not been in the public domain before — are as follows:
J. Z. Titow’s calculations of yields on the estates of the bishops of Winchester from 1211 to 1349.
The individual manorial accounts enrolled annually in the Pipe Rolls of the bishops of Winchester constitute the single largest corpus of yield information still extant. It is intimidating in its sheer magnitude. The Herculean task of calculating and tabulating over 9, 200 grain yields recorded for the forty to fifty manors on this estate — crop by crop, year by year, and manor by manor — from the earliest recorded harvest in 1211 until that gathered at the height of the Black Death in 1349, has been undertaken by Dr Jan Titow. His Winchester yields: a study in medieval agricultural productivity (Cambridge, 1972) has long constituted the single largest compendium of historical yield data in the public domain. With his permission, these yields have been converted into electronic format and incorporated into the Medieval Crop Yields Database, where they constitute its single largest component. There are some 8, 340 yield observations for the four principal grains of wheat, rye, barley, and oats and an additional 870 observations for such minority grains as winter barley and the grain mixtures, maslin (wheat and rye), mancorn (wheat and winter barley), and dredge (spring barley and oats).
D. L. Farmer’s calculations of yields on the estates of the bishops of Winchester from 1350 to 1453 (and extended by Bruce M. S.Campbell to 1471).
A further 7, 760 wheat, barley, and oats yields calculated by David Farmer (and augmented by Bruce M. S. Campbell for the period 1453-71 and those years missing from Farmer’s series due to the poor quality of some of the microfiche copies of the original rolls upon which relied) extend the Winchester yield data to the end of direct management in 1471 and almost double the number of yield observations available for this one estate. Farmer was only ever able to publish these yields as annual un-weighted averages (D. L. Farmer, ‘Grain yields on the Winchester manors in the later Middle Ages’, Economic History Review, 2nd series, 30 (1977), 555-66; D. L. Farmer, ‘Prices and wages, 1350-1500’, 431-525 in E. Miller, ed., The agrarian history of England and Wales, Volume III, 1348-1500 (Cambridge, 1991), 506-8). The original crop-, year-, and manor-specific calculations are, however, preserved in his papers at the University of Saskatchewan Archive and it is these that have been used here (with the permission of the U. of S. Archive)
For this one outstandingly well-documented estate over this 260-year period almost 17, 000 detailed grain-yield observations are now available [see ‘Listing of Winchester Pipe Rolls and yield data’]. These constitute over half of all the observations in the Medieval Crop Yields Database. It should be noted, however, that neither Titow nor Farmer collected yields of peas and beans and Farmer excluded rye, winter barley, maslin, beremancorn, and dredge from his post-1349 calculations. Yields of these crops therefore remain to be collected. A more serious limitation of the Winchester yield data is the many gaps that occur whenever there is no extant Pipe Roll, either because it was never compiled (either because the bishop had just died or because of a vacancy) or has not survived. Only for the Winchester manor of Bishops Waltham after 1350 is it possible to fill some of these gaps using the original compotus rolls from which the Pipe Rolls were compiled and for which this manor alone retains a reasonably full series.
D. L. Farmer’s calculations of yields on the estates of the Abbey of Westminster from 1271 to 1415.
After the estates of the bishops of Winchester, the next greatest body of available yield data is that gathered by Farmer for a selection of the nineteen best-recorded manors belonging Westminster Abbey. These, too, were only ever published as averages for the estate as a whole (D. L. Farmer, ‘Grain yields on Westminster Abbey manors, 1271-1410’, Canadian Journal of History 18 (1983), 331-47). For the 144-year period from 1271 to 1415 this furnishes a further 4, 080 wheat, rye, barley, and oats observations and 880 observations for assorted other grains. Again, his detailed manor-level calculations are preserved in the David Farmer Papers at the University of Saskatchewan Archives and are used here with that archive’s permission (see ’Acknowledgements’). Although this is a less substantial and consistent body of data than that provided by the Winchester Pipe Rolls, because it is derived from individual compotus rolls rather than enrolled accounts, there is scarcely a year without at least some available yield information. The problem with the Westminster yields is therefore less the chronological gaps in the data than the data’s ever-changing volume and coverage and the relatively limited chronological span of the information.
Together, Farmer’s 12, 720 grain-yield calculations for the Winchester and Westminster estates, all of them hitherto unpublished and inaccessible, constitute the single greatest body of yield information gathered by any one scholar, exceeding by 3, 500 observations the number compiled by Titow. This is all the more remarkable given that Farmer undertook most of this work in Canada using microfilm and microfiche copies of the original documents.
Bruce M. S. Campbell’s calculations of yields in Norfolk and neighbouring parts of Suffolk from the 1260s to the 1460s.
The next largest body of yield data, likewise hitherto unpublished, is that compiled by Bruce M. S. Campbell — creator of the Medieval Crop Yields Database — for Norfolk from a comprehensive trawl of all extant manorial accounts for that richly-documented county (the manors and accounts are listed in Bruce M. S. Campbell, English seigniorial agriculture 1250-1450, Cambridge, 2000, 453-65). This includes long runs of yields for several manors just across the county boundary into Suffolk as well as yields for a handful of demesnes in the Suffolk Breckland published by Mark Bailey (Mark Bailey, A marginal economy? East-Anglian Breckland in the later Middle Ages, Cambridge, 1989, 101-3). Geographically this is by far the most sharply focused body of yield data even though the information relates to a heterogeneous collection of estates, lay as well as ecclesiastical and small as well as large. Even so, it is the estates of such religious institutions as Norwich Cathedral Priory, the abbey of Bury St Edmunds, and the bishoprics of Norwich and Ely that dominate the record. The 5, 100 grain-yields (4, 810 for wheat, rye, barley, and oats) span a 195–year period from 1265 to 1460 with very few years lacking information altogether but much variation in annual coverage.
Bruce M. S. Campbell’s calculations of yields on the estates of Canterbury Cathedral Priory from 1268 to 1390.
Specifically for this project, and with the assistance of additional funding from the British Academy, a further 3, 120 grain yields have been collected for the extensive and well-documented estates of Canterbury Cathedral Priory in Kent, Surrey, and Essex. Since there is no Winchester and only one Westminster manor in Kent, this material adds significantly to the overall geographical coverage of the database. The Canterbury material is particularly full for the period before 1350, it also amplifies the amount of yield information available for the crisis years of the first half of the fourteenth century, which are only patchily served by the Winchester, Westminster, and Norfolk datasets. Although the Canterbury Cathedral Priory beadles’ rolls (as the CCP manorial accounts are known) have been much studied, the wealth of yield data which they contain has never been systematically exploited and is here made available for the first time. A run of yields for the years 1280-9 for the royal manor of Ospringe in Kent have also been included in this exercise. Marilyn Livingstone assisted in the considerable task of transcribing the relevant information from the original rolls and inputting it into a spreadsheet.
Bruce M. S.Campbell and Marilyn Livingstone’s calculations of yields on the Hampshire manors of Froyle and Silkstead.
Also specifically for this project and to help bridge some of the gaps in the Winchester yield data two additional Hampshire yield series were compiled for the manor of Froyle belonging to the nunnery of St Mary’s at Winchester (by Campbell) and the manor of Silkstead belonging to St Swithun’s Cathedral Priory Winchester (by Livingstone). Both manors were immediately adjacent to properties of the bishops of Winchester — Froyle neighboured the bishopric manor of Bentley in north-east Hampshire and Silkstead faced the bishopric manor of Twyford across the Itchen valley just south of Winchester — and therefore serve as surrogate Winchester manors. Froyle is remarkable for its relatively complete early run of yields from 1236 to 1262 plus some later yields in the 1360s and 1370s (see ‘Sources’) and Silkstead for a steady drizzle of yields from 1275 to 1390 (see ‘Sources’). Together, they reinforce the point that, remarkable as are the yield data contained in the Winchester Pipe Rolls, building a continuous chronology of yields is only possible by drawing upon data preserved in other archives for other estates.
Christopher Whittick and Anne Drewery’s calculations of yields on the Battle Abbey manor of Alciston in East Sussex, 1336-1492.
Due to the widespread abandonment of direct demesne management during the final decades of the fourteenth and opening decades of the fifteenth centuries, yield series that span the fifteenth century are rarer even than those that span the thirteenth century. For that reason, with the aid of a Margery Grant from the Sussex Archaeological Society, Christopher Whittick and Anne Drewery were commissioned to reconstruct a yield series for the exceptionally well-documented manor of Alciston in East Sussex (the only Sussex manor of chronological significance in the database), where Battle Abbey maintained direct cultivation until at least the early 1490s. Due to their efforts, Alciston now boasts the fullest and longest run of fifteenth-century yields currently available, adding a further 360 yields to the database and extending the latter’s chronological scope to the final decade of the fifteenth century. Matching these fifteenth-century yields for Alciston with others spanning all or part of the same period is now a high priority (see 6 ‘Outstanding priorities and future plans’, below).
Yields calculated from published editions of manorial accounts.
Several published editions of manorial accounts have furnished some valuable short runs of yields [see ‘Sources’] amounting to 250 observations in total, notably for Sevenhampton, Wiltshire, 1276‑87; the estate of Bolton Priory in the West Riding of Yorkshire, 1297-1324; and Petworth and several satellite manors in the Weald of Sussex, 1347-53.
Published and unpublished yields calculated by other historians.
Finally, 1, 000 grain yields have been included from a range of published (and one unpublished) sources relating to an assortment of estates, including properties of the countess of Albemarle, Merton College, Crowland Abbey, and the bishops of Ely, Exeter, and Worcester [see ‘Sources’]. Although they represent barely 4 per cent of the yield observations in the database, they reinforce and extend its chronological and geographical coverage in a number of useful ways.
What proportion of the total number of medieval yield observations that might potentially be collected is represented by the combined total of 30, 000 observations in the Medieval Crop Yields Database, is impossible to establish exactly. There is no single comprehensive register of extant manorial accounts, other than the summary guide to manorial records provided by the Manorial Documents Register maintained by the National Register of Archives [http://www.nationalarchives.gov.uk/mdr/help/mdr/mdrfaq.htm]. Nor does the fact that accounts survive mean that they necessarily relate to demesnes that were managed directly and therefore contain information on the crops sown and harvested. Usually only an inspection of the documents will establish whether that was the case. In the fifteenth century, when there are many surviving series of accounts but few for demesnes kept in hand, finding those with usable data requires patience and perseverance.
What is plain, is that a great deal of useful material remains to be identified and collected both in the major public archives and in those maintained privately. Of the latter, the more important collections not as yet tapped include those of the Berkeley estate at Berkeley Castle, Durham Cathedral Priory, Glastonbury Abbey at Longleat House, and a number of Oxford and Cambridge colleges. There is no doubt that incorporating additional material from these and other archives would greatly amplify the geographical and chronological scope and coverage of the database and help bring the results derived from it into sharper focus. It is equally clear that the Medieval Crop Yields Database already contains a clear majority of the yield observations that might eventually be collected because it incorporates most of the information from the greatest and best-preserved manorial archives of them all. Possibly the 30, 000 yield observations contained in the database represent about three-quarters of all the material that might be collected.
How accurate and reliable is the yield information contained in the database given the varied provenance of the observations?
Potential sources of noise and error are several:
- The medieval clerks themselves occasionally made errors, especially when enrolling data from the original compotus rolls onto either fair-copy or enrolled accounts (which is why the original compotus rolls, rough and scrappy though they may be, are always to be preferred when they exist). These mistakes have silently been corrected whenever they have been spotted.
- Transcribing data from these accounts is likely to have introduced further errors, especially when the documents are abraded, damaged, or in some way idiosyncratic. The possibility that some entries have been misread or misunderstood cannot be excluded. Not all historians will have been equally scrupulous in dealing with ‘old grain’ left over from the previous harvest; ‘new grain’ threshed and accounted for between harvest and the end of the accounting year in Michaelmas; and the complications that arise from the use of heaped versus razed (i.e. struck) bushels.
- The dating of the original documents can also pose problems, especially when abbatical and episcopal years are used (although regnal years are not without their ambiguities). The dates given in calendars and catalogues are not always to be trusted. A good check upon whether accounts are genuinely consecutive is whether the livestock numbers at the Michaelmas terminating the earlier account exactly match those enumerated at the opening Michaelmas of the following account. All independent information that might be used to clinch the date of an account should be noted. Obviously, a correct yield entered against the wrong year is a source of misinformation. Fortunately, systematic misdating can usually be spotted and rectified, as in the case of Farmer’s dating of the post-1349 Winchester yields by accounting (i.e. the year in which the harvest was reported) rather than harvest year (i.e. the year in which the crop was grown and harvested).
- Finally, inputting error may have occurred when the yield information was entered into the database. This is mostly likely when the transcriptions and calculations in question were the work of another historian, as most particularly with some of the un-tabulated transcriptions of David Farmer.
To minimise error, great care has been taken at all stages in the creation of the database, especially when the original data were in a hand-written manuscript format. All the most problematic material was carefully checked. The sheer magnitude of the task has, however, made it impossible to check everything. Instead, the database has been filtered for results that are significantly out of line with the normal distribution of yields. When in doubt, yields of uncertain accuracy have been eliminated. A ‘Wiki’ facility has also been incorporated into the database so that any errors spotted by users of the resource can be reported and, in due course, corrected. The database itself has already been subjected to a great many analyses, each of which has provided an opportunity for filtering, checking, and correcting the raw data. The fact that analyses run on independent subsets of the data have yielded broadly similar results and chronologies is also regarded as evidence that the database is internally consistent and reliable. Anyone who wishes to make use of the detailed information contained in the database (e.g. for a particular manor, year, or crop) is, however, strongly advised to check this information against that contained in the original documents (as identified under ‘Sources’).
Why does the database focus so exclusively upon yields per seed?
The rates at which seed was sown, the choice of crops that were cultivated, and the length of rotations were all material to the productivity of arable land. Nevertheless, the database focuses exclusively upon the rate of return upon the seed sown (known as the ‘yield per seed’ or ‘yield ratio’) even though this is a partial and, indeed, can be a misleading measure of land productivity. For instance, it does not necessarily follow that because yields per seed were high, yields per acre were also high — and vice versa. The focus upon yields per seed is for four reasons:
1. | Information on the quantities of seed sown and grain harvested are more consistently recorded than any of the other components of productivity, including the acreages sown and seed sown per acre (which are not given in all accounts). |
2. | Because the yield per seed is a ratio it is immune to the distortions to which absolute measures (such as the yield per acre) are susceptible, when and wherever customary bushels and/or acres were in use. |
3. | The yield per seed highlights year on year variations in harvest quality more effectively than the rate at which a crop yielded per acre (which can be influenced by changes in the seeding rate). Establishing annual and long-term variations in growing conditions for grain was one of the principal reasons why the Medieval Crop Yields Database was created in the first place. The rate of reproduction of grain, for instance, can be correlated with other proxy measures of environmental conditions, such as the growth record of trees, the chemical composition of Greenland ice cores, and the comments of contemporaries on meteorological conditions. |
4. | The yield per seed has long been the preferred productivity measure of historians, with the result that much more information is currently available than for, say, yields per acre. David Farmer, for example, who collected more yields than any other historian, concentrated almost exclusively upon yields per seed. Jan Titow, in contrast, collected and published yields both per seed and per acre, but had far more of the former than the latter. |
The raw yield figures contained in the database are those computed by the historians who gathered them. All are gross of seed and the vast majority are net of tithe, which was usually deducted at harvest in the field. The exceptions are when other arrangements were made for payment of tithe or, when the same landlord owned both the demesne and the tithes (and therefore did not pay himself tithes from the demesne). Such cases are not easily detected and consequently are not identified in the database.
The approach adopted with the chronologies reconstructed from these data is different. First, in order to reconstruct the total harvest inclusive rather than exclusive of tithes, the ‘tenth’ of the original harvest deducted as tithe has been reinstated. Second, since, as E. A. Wrigley has argued convincingly (‘Some reflections on corn yields and prices in pre-industrial economies’, 92-130 in People, cities and wealth: the transformation of traditional society, Oxford, 1987) it was the yield net of seed that determined the relative supply and price of available food, all chronologies have been expressed net of seed. The formula used to make this double adjustment is [(Yield per Seed ÷ 0.9) – 1]. A simple mathematical procedure will therefore reverse either or both adjustments.
How representative are these demesne yields likely to be of yields in general?
It might be objected that the yields obtained on the large demesne farms of lords (which, collectively, can have accounted for no more than 30 per cent of all tillage bearing crops) are hardly representative of arable production as a whole. Demesne producers, of course, enjoyed greater command over land and capital than the country’s more numerous occupying tenantry. They also farmed on a far greater scale and often with different objectives in terms of the relative primacy accorded production for sale versus consumption. Certainly, such great and perpetual institutional producers as the cathedral priories of Canterbury, Ely, Norwich, Winchester, and Worcester, or the abbeys of Battle, Crowland, Ramsey, St Edmund’s, and Westminster, were better able to weather economic and environmental hazards than the plethora of petty producers who must have cultivated well over two-thirds of all the land in the county and for whom a major back-to-back harvest failure or heavy mortality of cattle or sheep may well have spelled economic ruin. Indeed, great lords may actually have profited from the high prices which were the invariable accompaniment of such difficulties.
Yet the documented experience of the demesnes is not wholly unrepresentative of agriculture within the country. Lords and peasants shared a common labour force, know-how, and agricultural technology. They supplied much the same markets and were informed by the same prices. In fact, it made little economic sense for lords to manage their demesne lands directly, unless they were able to obtain at least as good a return as from leasing them for competitive rents to tenants. Above all, the soils they cultivated were similar and pests, parasites, pathogens, and the weather did not discriminate between them, especially when their lands lay intermingled. Whether lords and tenants obtained the same level of yield per seed and per acre is a moot point which continues to be much debated, but there seems every reason to suppose that the variations in harvest quality experienced by lords were shared by their tenants, with the results that the yields recorded across a wide sample of demesnes may be regarded as diagnostic of those experienced by most agricultural producers within that region.
If variations in demesne yields are broadly representative of yields in general there should be a correlation between yields and prices, since the latter reflected the total supply of grain to the market not merely that produced on demesnes.
In the case of wheat, the most commercialised of grains, there was a positive correlation between the percentage wheat price and the percentage grain price (inverted and advanced one year) of +0.54 across the 180 years 1260-1439 and +0.60 across the 100 years 1300-99, when data coverage is fullest. In fact, between 1310 and 1330 and between 1380 and 1400 the correlation between wheat yield and price reached and exceeded +0.80. Such strong correlations suggest that across the period as a whole, and for much of the fourteenth century in particular, annual variations in demesne wheat yields are broadly representative of wheat yields in general. The correlations between composite grain yields and composite grain prices over the same period, and especially during the first half of the fourteenth century, support much the same conclusion. Respective correlation coefficients for the years 1260-1439 and 1300-1399 are +0.54 and +0.60.
Further evidence of the representativeness of the demesne yield evidence is furnished by a comparison of annual variations in yields and tithe receipts, using tithe series recently assembled for the estates of Durham Cathedral Priory in the north-east of England (B. Dodds, ‘Estimating arable output using Durham Priory tithe receipts, 1341-1450’, Economic History Review, 2nd series, 57(2004), 245-85) and the Hampshire parish of Hambledon in the south of England (ESRC project, RES‑000-23-0005, ‘Tithe income and management in southern England 1280-1480’, directed by R. M. Smith, R. H. Britnell, and B. Dodds) — I am grateful to Dr Dodds for making these materials available. The significance of this comparison lies in the fact that tithes were paid by all grain producers and should therefore reflect the general quality of the grain harvest.
In the case of national yields and Durham cash tithe receipts, the comparison is between two entirely independent datasets for two geographically separate sources of data. Even so, from the 1340s to the 1390s the correlation between the two series varied between +0.40 and +0.75. As is to be expected, the correlation between Wessex grain yields and receipts of grain tithes at Hambledon in Hampshire is closer and from the 1330s to the 1380s was +0.75 of better, with an overall correlation coefficient of +0.78.
Such high correlations lend strong support to the proposition that, at least for the period when the yield data are most abundant, the yield variations registered by demesnes are representative of yield variations experienced by all grain producers. The fall off in the correlations from the end of the fourteenth century is striking and may either be a product of the shrinking number of demesnes represented in the Medieval Crop Yields Database, or because from this time the tithe payments became more formalised and cease to be such a direct function of the prevailing harvest. This is an issue worth closer investigation.