Introduction

This guide is to keys for locking and unlocking locks and padlocks. Locks and keys are symbolic of personal property, control and exclusion, so they are interesting items to record and study.

Most of the keys recorded on the PAS database are made from copper alloy, but most excavated keys (and probably most keys in use in the past) are made from iron. Keys of the two materials seem to be surprisingly different, and we should not assume that what holds for iron keys is also true for copper alloy keys.

The best source for Roman keys covers only the iron examples (Manning 1985) but also mentions a few copper-alloy examples in passing.

The best source for early-medieval copper-alloy keys is Megan von Ackermann’s PhD at the University of York, which will shortly be available on Ethos. There is also an excellent short section on early-medieval iron keys in the Flixborough report (Ottaway in Evans and Loveluck 2009, 187-195).

The best sources for early-medieval and medieval keys are Goodall in Biddle 1990, Ward Perkins 1940, and Egan 1998. There are no particularly good sources for post-medieval keys.

Different types of locks and keys

There are several different types of lock, which use a sliding bolt with tumblers, levers or springs to keep the bolt from moving. They do not necessarily all require different forms of key; the same key form can be used to operate different kinds of lock.

Terminology can be inconsistent, and occasionally the same term is used for different things. For example, the term ‘slide key’ has been used to refer to a key which slides along the springs of a padlock, squeezing them together so that the padlock can be opened (Ottaway and Rogers 2002, 2866-2869). But ‘slide key’ is also used to refer to a key which lifts tumblers out of a perforated lock-bolt and then slides the bolt open (Manning 1985, 92-3); and this type of key is also sometimes called a ‘lift key’. Rotary keys all turn, but can then lift tumblers or levers, or push springs, so that the lock can be opened.

It will be clear from this that any jargon terms such as lift key or rotary key cannot be used without full explanation of what you mean, and are no substitute for a meticulous description.

The word ‘key’ can also be used for a tool to grasp and turn something other than a lock; perhaps to open a gas meter cabinet (a gas key) or to wind a clock (a clock key). We use the term KEY (WINDING) for these. Some keys open a valve, but the equivalent part of a tap is called a ‘handle’; see the guide to taps for further details.

PAS object type(s) to be used

KEY (LOCKING) should be used for all keys that work with locks or padlocks. There are many alternative terms available on the mda thesaurus, but as the different terms are poorly defined and inconsistently used, they should be avoided. Moreover, in practice we cannot normally distinguish fragments of slide keys, lift keys and rotary keys, or tell keys from mounted locks from those used with padlocks. What we can see is the overall shape, and so we classify keys according to shape rather than specific function.

For keys which simply turn round and round, rather than lock and unlock, use KEY (WINDING). Even though not all of these keys will be used for winding clockwork mechanisms, there is no alternative on the mda thesaurus.

For key-like items which turn to open and close a tap, use TAP.

PAS classifications and sub-classifications to be used

For T-shaped or L-shaped keys, please add ‘T shaped’ or ‘L shaped’ to the classification field.

There is no generally accepted typology for Roman copper-alloy keys or fragments, but we can flag up the major divisions. For an openwork handle (normally with three lobes) please add ‘openwork handle’ to the classification field. For a solid rectangular handle with loop, please add ‘solid rectangular handle’ to the classification field. For a handle in the shape of an animal, please add ‘zoomorphic handle’ to the classification field. For ring-keys, please add ‘ring key’ to the classification field. For anything else, or uncertain fragments, leave the classification field blank.

For early-medieval and medieval keys for mounted locks, the Winchester types developed by Goodall (in Biddle 1990) seem to be the most useful at present. Please add them, if you can, to the classification field in the following format: Winchester type 2. Be careful to use Arabic numerals for Winchester types.

The exception is for London type VI keys (Ward Perkins 1940), which do not have an equivalent in the Winchester typology. Be careful to use Roman numerals for the London typology. Add London types in the following format: London type VI.

Casket keys do not fall into either Winchester or London types. For these, it is useful to use ‘casket’ in the Description field.

For padlock keys, please again use the Winchester types (Winchester type A, Winchester type B, Winchester type C). For the mysterious variety with T-shaped extension beyond the bit, use ‘type C variant’ in the classification field.

Terms to use in the Description

In general we follow Goodall’s terminology, which is largely also followed by Egan (although Egan uses the term ‘shank’ rather than ‘stem’).

This tripartite division does not work particularly well for some Roman keys, especially those with a suspension loop, solid rectangular element, stem and bit; it can be very hard to work out whether the solid rectangular element is part of the handle, or part of the stem. For Roman keys, it is therefore perfectly acceptable to use the term ‘handle’ instead of bow.

Roman slide keys have ‘teeth’ on their bits (WAW-9D3A02 illustrated above has six teeth) whereas the word ‘ward’ seems to be more commonly used for other keys.

Similarly, if the handle end of a medieval padlock key is very unlike a rotary key’s bow, call it a handle.

Latchlifters

The term ‘latchlifter’ is often applied to simple hooks with a handle which could be used to open a bolt. Manning (1985, 88-89, pls. 37-40) lists several iron examples from late Iron Age and Roman sites; they also occur in the early-medieval world, up to the 9th century (Ottaway in Evans and Loveluck 2009, 190). They do not appear to have been made from copper alloy. We have very few recorded on the PAS database; if you do record one, use the object type LATCHLIFTER. A date-range of 100 BC to 800 AD is probably appropriate.

T-shaped and L-shaped keys

T-shaped or L-shaped keys were used with Roman tumbler locks (Manning 1985, 90-92, pl. 40) and were mostly made from iron. They occasionally occur in early Anglo-Saxon graves, and there are examples from Flixborough. Ottaway suggests that they went out of use at the start of the 10th century (in Evans and Loveluck 2009, 190). A date of c. 43-900 AD is probably sensible for the iron examples.

We have a few copper-alloy examples recorded on the PAS database, which are probably of similar date; and, exceptionally, one 95% silver example (HAMP-66BB77). Some are decorated, and where the decoration is diagnostic, it usually suggests a Roman date. Note the Roman-style handles on HAMP-66BB77 and BH-0852F6.

T-shaped keys should not be confused with early Anglo-Saxon girdle-hangers, which are superficially similar to copper-alloy T-shaped keys, but flat in cross-section, non-functional and generally decorated with stamps.

Roman keys

The earliest keys in Britain date to the Roman period. They are usually made in two pieces, with an iron stem and bit, and a copper-alloy handle or bow, but occasionally we record complete copper-alloy examples.

There is no generally accepted typology for Roman copper-alloy keys or fragments, but there are three distinctive shapes of handle; openwork (normally trilobate) and solid (normally rectangular, with a loop) are the commonest, and there are also a small number of zoomorphic handles. There may be many other less distinctive Roman keys too.

Openwork handles, normally three-lobed

These usually consist of the handle or bow only, because the stem and bit were made from iron. Exceptionally, BH-57A610 is made entirely from copper alloy, and has a hollow stem and massive bit which together look horribly like a medieval rotary key of London type VI (see below); caution should therefore be exercised when dealing with fragments. SOMDOR-9B8C55 is similar, but with an iron stem and bit surviving.

These handles can also be found on smaller keys, and in a variety of patterns, not all of which have the three lobes (see below for examples). There are also some quatrefoil or quadrilobate key fragments recorded on the PAS database as Roman, but as all complete examples are medieval, a Roman date is less likely to be correct. If you are unsure whether to go for a Roman or medieval date, look closely at the collar (if it survives) as Roman collars tend to have multiple transverse grooves and narrow ridges, and medieval collars tend to have a larger central rounded component.

Some openwork handles have separately made iron stems and bits (see illustration above) but others, mainly smaller, have integral copper-alloy stems and bits (see illustration below). Note that in some cases (such as SF-491826 on the left below) the bit projects at right angles to the handle.

Because of the variety of shapes, it is best to add ‘openwork’ to the classification field. The term ‘trilobate’ is often used in the Description field, and seems more popular than the alternatives (which include ‘trefoil’ and ‘tripartite’, but not ‘trilobite’ because a trilobite is a fossil arthropod living about 500-250 million years ago).

Nina Crummy quotes a date of after 150 AD for these (Crummy 1983, 126, citing Der Obergermanisch-Raetische Limesdes Römerreiches 8, Taf 12,51).

Solid rectangular handles

These keys have a solid, thickened rectangular area at one end of the stem, forming a handle; it normally ends in a loop, usually circular or lozenge-shaped but occasionally of other shapes. They appear to date from the entire Roman period, and can have bits of various shapes.

With most, the bit projects in the same plane as the bow, like a modern key, but the teeth of the bit then project at a right angle, either forwards or sideways. These teeth are thought to have fitted into the holes on openwork ‘lock bolts’ such as HESH-058774 (see LOCKS for more details on lock bolts) and the key then used to slide the bolt across. A key and its corresponding lock-bolt are in the BM collection, at 1912,1021.2. The keyholes for these keys would have been L-shaped.

Those shown below are all of Manning’s type 2 slide keys (Manning 1985, 93); examples of Manning type 1 (with a longer, L-shaped or curved bit) have apparently not yet been recorded on the PAS database.

Other examples of keys with solid handles have a simpler bit which projects at right angles to the handle. These may have been rotary keys.

Many of these solid rectangular handles, like the openwork handles, will have had iron stems and bits. These occasionally survive (as on LANCUM-6B5B80) but it is more common to find the copper-alloy handle on its own.

Keys with the bit at right angles to the plane of the handle

Where the bow is turned at 90˚ to the modern orientation, there is a strong likelihood that the key is Roman. This feature is occasionally found with openwork key handles (such as SF-491826 illustrated above) and with solid rectangular handles (such as BH-E374F2, LVPL-556548 and SUR-19B512, all illustrated above).

It is also found with less diagnostic bow forms, such as those shown below. 

Zoomorphic handles

Solid key handles are occasionally made in the shape of animals, usually showing the head, forelegs or more of a crouching lion. The most elaborate example on the PAS database is SWYOR-F1D5D6, asymmetrically modelled with prey (a ram’s head) in the lion’s mouth. A more representative group is shown below, with the most complete example (NLM-1E9032) retaining its iron stem and bit.

The PAS database also includes examples of other copper-alloy handles in the form of animals, usually identified as being a bear, dog or wolf. These are sometimes recorded as key handles (e.g. SOM-B0C429 and BUC-BDD866) and sometimes as knife handles (e.g. NCL-864495, SF3834 and LVPL-1DB33C). Two complete examples of knives with animal-head handles are in the BM collection (1856,0701.1132  and 1851,0813.45). Other functions are no doubt also possible.

Ring-keys

Ring-keys combine a finger-ring and a key, and are designed to be worn on the finger. The primary function of ring-keys is as tiny keys for locking and unlocking, rather than as personal decoration, so although a few are so small as to be possibly non-functional, please use KEY (LOCKING) for all of them. Add ‘ring key’ to the classification field, and the Guiraud type (e.g. ‘Guiraud type 5a’) to the sub-classification field.

The finger-ring part is usually quite simple, with little decoration. The stem runs along the finger, and the bit also lies flat along the finger; the ring would clearly have had to be removed from the finger to use it. The junction between stem and hoop is usually small and fragile and can often break, leaving an easily-overlooked scar.

Ring-keys were classified by Guiraud (1989, 191-3) as her finger-ring type 5. Guiraud divided them into five sub-types on the basis of the bit, but only Guiraud types 5a and 5b are at all common. Type 5a usually has a hollow stem and a bit with a slot immediately below the stem, but can come in simpler versions as well (e.g. WILT-C592D7, shown below). Type 5b has a very short stem and a square or rounded bit with openwork centre.

The other types are very rare; type 5c has a double-sided bit, type 5d appears to be a cruder version of type 5a, and type 5e has a decorative projection which does not appear to have worked as a functional key.

If the bit is missing, in theory it is impossible to tell which of the sub-types of ring-key you are dealing with, but in fact almost all type 5a keys have the stem projecting from one end of the bezel, and almost all type 5b keys have the stem projecting from the centre of the bezel.

Crummy (1983, 84) suggests that type 5b (with ‘flat wards’) may be early Roman, whereas type 5a may be 3rd or 4th century in date.

Note that it is not easy to photograph ring-keys so that the details are clear. Firstly, check the depth of field and make sure that all parts are in focus. Secondly, be careful that you have projected (arranged) your different views properly. Thirdly, as the stems and bits of a ring-key must lie along the finger so that it is comfortable to wear, they can be difficult to see in a normal record photo, so as with other finger-rings, it is often useful to include an oblique view (see below) as well as the normal record shots.

Other Roman keys

There are also some small casket keys which can be identified as Roman keys from the distinctive form of their bit. It is of the same type as that found on keys with solid rectangular handles (see above); with sideways-projecting teeth which are thought to have fitted into the holes on openwork ‘lock bolts’. Manning calls these ‘slide keys’. Examples include NARC-08C7B1, HESH-8BB970, SUSS-DB3897, ESS-857F54,  WMID-E806A1, SUR-7EA641 and LVPL-9F4D2E.

A key which has a small simple L-shaped bit (instead of the substantial toothed bit) is more likely to be early-medieval, such as IOW-A346E4 (illustrated below).

Early-medieval keys

Early and middle early-medieval keys

Early Anglo-Saxon graves occasionally contain latch-lifters, and even more occasionally can contain T-shaped or L-shaped keys (see above for both of these categories). Oddly, 7th-century graves do occasionally contain padlocks, normally with boxes (e.g. Harford Farm 7 and 18, Finglesham 6, Didcot 2, Castledyke 1) , but the keys used for these padlocks appear to be absent or unrecognised.

Although Winchester type 2 keys are known from several classic middle Anglo-Saxon sites (e.g. Hamwic, Wicken Bonhunt, Thwing and Brandon) none comes from a context that definitely pre-dates the 9th century.

Late early-medieval keys

Keys come into use in the 9th century at York, Flixborough and Winchester (Rogers 1993, 1420; Evans and Loveluck 2009, 194; Goodall in Biddle 1990, 1007). By the late 10th century, keys were well known enough to feature as the (probable) solution to the Exeter Book’s Riddle 44. This gives us a possible clue to the appearance of the early-medieval key, in that the object concerned has a hole in the front (foran is þyrel).

The characteristics of early-medieval keys are: a drop-shaped bow, tapering into the stem; an L-section bit; and a suspension loop at the far end of the bow. Any one of these is diagnostic of an early-medieval date. Some also have decoration in the Winchester style (e.g. WILT-C520E4 and DOR-6E063A), which helps to assign a precise date in the 10th century for these particular examples.

Winchester types 1 and 2

Two of Goodall’s Winchester types are early-medieval in date, types 1 and 2, defined as having L-section bits. Type 1 has a projecting stem-tip, and is by far the commoner type among the iron keys at Winchester. Type 2 has a hollow tip to the stem, and is much commoner among the copper-alloy keys in the PAS assemblage. There is an explanation of how type 1 and 2 keys work in Biddle 1990, 1016-19.

It seems possible that type 2 is the earlier form, as it is found in a mid to late 9th-century context at Flixborough, and without other dating evidence at Brandon, Hamwic, Whitby, Wicken Bonhunt, Thwing and many other classic middle to late Anglo-Saxon sites (Evans and Loveluck 2009, 194-5; Hinton 1996, 50-1; Tester et al 2014, 182-3). The end date of type 2 is less certain, but c. 800-1100 is probably fairly safe.

At Winchester, iron keys of type 1 are found in contexts of the early 10th to mid 12th century. c. 900-1150 AD. This date seems likely for copper-alloy examples too (see the example illustrated above with 10th-century Winchester style, DOR-6E063A).

It is not easy to photograph the salient features of these keys, but do try to show the cross-section of the bit and, if possible, any hollow end.

The earliest casket keys

Casket keys are diminutive, simple keys of about 35-40mm in length, and are generally hard to date. The small keys known as ‘casket keys’ also seem to have begun in the 9th century. One (possibly incomplete) copper-alloy example was found at the 7th- to 9th-century site at Brandon, Suffolk (Riddler in Tester et al 2014, 183, no. 8331). A few others have diagnostic early-medieval features, such as a drop-shaped bow, suspension loop or L-section bit. One with L-section bit is illustrated above (IOW-A346E4).

But as these small, undiagnostic keys continue to the end of the medieval period, it is impossible to date unstratified, undiagnostic casket keys this early. See below for more on casket keys, which we tend to date to c. 1100-1500 AD.

Medieval keys

It is still true that “the close dating of medieval keys is a matter of great difficulty” (Ward Perkins 1940, 133) and this is made harder by the fact that most excavated keys with secure dates are made from iron, and are rather different from the copper-alloy examples that make up 90% of the medieval keys on the PAS database.

The main typologies are Ward Perkins’s, developed for the keys in the London Museum; and Goodall’s, a modification of the London system developed in Winchester (Biddle 1990, 1005-36). The London system uses Roman numerals, and the Winchester system uses Arabic numerals.

Both typologies are based on iron keys, although there are occasional examples in copper alloy. London type VI keys are the exception, being exclusively of copper alloy.

What most medieval copper-alloy keys look like

To save you reading the whole of this guide, most medieval copper-alloy keys fall into either London type VI (large copper-alloy keys with hollow stems) or are ‘casket’ keys of various types (some of which fall into Winchester type 9). This guide will look at these two types first, and then go on to describe a few other keys. The other main Winchester types (types 3-8) are mainly made from iron and only very occasionally occur in copper alloy.

‘Casket’ keys

Casket keys are the commonest form of medieval key on the PAS database. They are small and simple, normally 35-40mm long; if your key is over 50mm, it is unlikely to be a casket key. The name was coined by Ward Perkins (1940, 144) but they were probably used on all sorts of small locks, on cupboards, chests, cases, etc.

Although almost all of the casket keys on the PAS database are of copper alloy, we know from excavated material that they were often made from iron as well, and are very similar in both materials.

The bow is normally simple and circular, but can occasionally be more elaborate; lozengiform bows are sometimes found. The stem is normally hollow at the end, but occasionally can be solid, tapering to project beyond the bit.

The bit is always quite simple, often symmetrical, sometimes uncut. They are normally made from cast copper-alloy, but can occasionally be made from folded sheet.

A group of at least 32 ‘casket’ keys, both copper alloy and iron, was found in a context of c. 1233-1280 at St Mary Spital in Spitalfields, London, and thought to be keys from patients’ lockers (Thomas, Sloane and Phillpotts 1997, 34-5, 202-3).

The start and end dates of casket keys are hard to pin down. They may start in the early-medieval period (see above, from Flixborough) but they are more common in medieval contexts. They certainly continue into the 15th century (from York, Ottaway and Rogers 2002, no. 14290; and Winchester, Biddle 1990 no. 3851). Until we have further evidence, a wide date-range of c. 1100-1500 AD seems sensible.

London type VI copper-alloy keys

The other common type of copper-alloy key recorded on the PAS database is a large key known as the London type VI key.

When Ward Perkins (1940, 134-141) established his typology of large London keys, it was the only type that was mainly found in copper alloy. It does not occur as a separate type in Goodall’s Winchester typology, presumably because none were found at Winchester.

Type VI keys are mostly 80-100mm long and around 8-10mm thick, of chunky proportions, with hollow ends to the stem (or completely hollow stems, as on IOW-A54593) and with massive, often complex bits. A few can be slightly smaller, but still with the same proportions. Note that BH-57A610, a Roman key with openwork trilobate bow (pictured above) has a similar hollow stem and massive bit, so exercise caution when recording fragments.

The bows are usually lozengiform or quatrefoil, sometimes circular. They can just consist of a frame, but often also have openwork designs internally or knops and mouldings externally.

Those with openwork bows often also have suspension loops, normally short and tubular; these may have been used with rings, chains or swivels but, if so, none survive in place. One key, WMID-25B5B4, has a leather thong surviving, but it is wrapped around the bow and does not actually use the ‘suspension’ loop.

Some of these keys appear to be regionally distinctive. Rogerson and Ashley (2012) have suggested that one type (image above, centre top) may have been produced in the Aylsham area of Norfolk; and another (image above, centre below), with lozenge bow and four circular perforations, seems to be common in the midlands, centred around Lichfield. A detailed study of the different designs of type VI keys, and their distributions, is long overdue.

Ward Perkins believed that type VI dated to the 14th and 15th centuries (1940, 140), but reliable dating evidence is scarce. A fragment from London, missing its bow, was found in a context of c. 1270-1350 (Egan 1998, no. 313).

Goodall dates lozenge bows found on iron keys at Winchester to no later than the 14th century (in Biddle 1990, 1007), and it is possible (but not certain) that this holds true for copper-alloy keys too.

The function of London type VI keys is uncertain. Egan (1998, 111) appears to agree with Ward Perkins that a hollow-ended stem is not appropriate for a door that needs to be locked from either side, and so is more likely to have been used for a cupboard or chest. As these keys are quite big, though, the lock will have been correspondingly large.

Other (less common) medieval keys

There are some smaller, not VI keys; LEIC-B955CA, DENO-7B5DBA with simple bits, generally. 50-70mm

GAT-954187 is like a type VI, but has a projecting stem. Not deeply split, so probably not really a type 5.

Winchester type 3 and type 4 keys (London type II and III)

These categories are really only relevant for iron keys, which are forged rather than cast (copper-alloy keys tend to be simply cast in one piece, or occasionally made from folded sheet).

Both Winchester type 3 and type 4 keys have hollow ends to the stem, but are defined by the way in which they are forged. Winchester type 3 keys (London type II) are rolled out of a single piece of metal; in Winchester type 4 keys (London type III), the bit is made separately and welded on.

Winchester type 5 keys – with deeply split stem

These are uncommon but very distinctive keys which are found in both copper alloy and iron, in a variety of sizes. They can be recognised by their deeply split stem.

Exactly how deeply the stem needs to be split to qualify it as a type 5, rather than just a key with a solid stem (as below), is not yet defined. As there are so few of this type, it’s best to add ‘Winchester type 5’ to the classification field even if the identification is a bit uncertain.

At Winchester, these are found in contexts of the 11th to 13th centuries.

Keys with solid stems

The remaining types are those with the solid stems, either in line with the end of the bit or projecting beyond. If larger than a casket key, these theoretically fall into Winchester types 6, 7, and 8, and London types IV, V, VII and VIII. But examples in copper alloy are rare, and it is not worth trying to allocate them to a typology which was essentially developed for iron keys.

There will always be occasional other oddities, such as HESH-6CD947 and ESS-24D08C, which appear to have had solid stems projecting beyond a pair of bits, one above and one below the stem.

Padlock keys

Different types of padlock can be unlocked using a variety of keys, some of which cannot be distinguished from keys for mounted locks. Because of this, the term KEY (LOCKING) should be used for any key, whether from a padlock or a mounted lock.

There is, however, one type of KEY (LOCKING) which is unique to the padlock, and on the PAS database we call this (in the Description field) a ‘padlock key’. It has a perforated bit which slides along the barbs of the padlock’s spring, squeezing them together so that the two halves of the padlock can be pulled apart.

Roman padlock keys

Iron padlock keys are fairly common in the Roman world. They are usually made from a single gently tapering strip of iron, with the narrower end turned over to form a small bow, and the wider end bent at right angles to form the perforated bit. Keys of this type were the commonest type of key found at Gorhambury Roman villa (Neal et al 1990, 147-8).

Manning (1985, 96-7 nos. O71-O74) includes three complete iron padlock keys (including 1853,0514.19 and 1877,0116.16) and one very large detached handle (1882,0206.17).

Padlock keys do not appear to have been made in copper alloy during the Roman period, and no padlock keys have been identified as Roman on the PAS database.

Late early-medieval and medieval padlock keys

Three types of padlock key were identified by Ian Goodall from the (exclusively iron) keys from Winchester. These were type A (bit set laterally to stem), type B (bit set centrally to stem), and type C (bit and stem in line). Type C includes Egan’s ‘shield shaped’ keys, which were Ward Perkins’s London type IX (Ward Perkins 1940, 135 and 143-4; Egan 1998, 102-3).

The Winchester type can go in the Classification field, and the word ‘padlock’ can be used in the Description field.

The terminology of padlock keys should mirror that of other keys as much as possible.  Hence they have bits and stems, and may have bows; if, however, the handle end is very unlike a normal key bow, call it a handle.

Winchester type A

Goodall defines these as having the bit ‘set laterally’ to the stem; in other words, the bit is attached to the stem by its edge. Type A keys are not particularly common, and are given a wide date-range by Goodall from early-medieval to post-medieval; the examples from Winchester (all of iron) date to the 10th to 15th centuries (Goodall in Biddle 1990, 1006).

The best-photographed example on the PAS database is shown below, but this does not have an end-on view showing details of the bit. It is always useful to take as many angles as you can, as these keys can be difficult to reconstruct from limited views.

Winchester type B

Goodall defines these as having the bit set centrally to the stem, and Ward Perkins calls them the ‘symmetrical form’. There are perhaps three main forms on the database, one with a lozenge bow, another with a circular bow with off-centre knop, and a plainer one with simple circular bow.

Ward Perkins dates this type to the 11th century at the latest (1940, 149-150), and Ian Goodall dates them to the 12th or 13th centuries (in Biddle 1990, 1006). Both were probably again working from iron examples. Copper-alloy keys with simple circular bows and bits set centrally to the stem are now known from a 12th- or 13th-century deposit at Coppergate, York (Ottaway and Rogers 2002, no. 12878), and 15th- or 16th-century contexts in Norwich (Margeson 1993, 162; no. 1312) and Exeter (Alison Goodall in Allan 1984, no. 183). A long date-range of c. 1100 to c. 1500 (or even longer) is therefore probably sensible for the type B, until we find more well-dated examples.

Winchester type C

Goodall defines this type as having ‘bit and stem in line’ (in Biddle 1990, nos 3725-30). The Winchester type C keys are all of iron, and have openwork rectangular or circular bits; they belong to the late 11th to possibly the early 13th century at Winchester, and are also found in mainly 12th-century contexts in London (Egan 1998, 100-102, nos. 267-8). There are not many known of copper alloy.

In the mid 14th century these keys seem to be superseded by a well-made group with ‘shield-shaped bit’ (Egan 1998, 102-3, nos. 269-272); this group also forms Ward Perkins’s London type IX. These keys may have continued being made in small numbers into the 19th century (see below) as ‘French night latch’ keys, so are difficult to date precisely.

Other possible padlock keys – with double or triple stem, and type C variant

Egan (1998, 103, no. 273) illustrates a key with double stem and bits set laterally to each half, from a late 14th-century context. It is paralleled to a certain extent by triple-stem examples, two from 12th- and 13th-century Gotland (MacGregor 1997, 66-67) and one from Icklingham, Suffolk (West 1998, fig. 55.8). We have a single example of this type on the PAS database, with double stem, LANCUM-B3B2F7.

Another object which appears to be a padlock key has never been found in an archaeological context, although we have nearly 30 now recorded on the PAS database, with no obvious geographical concentrations. We are currently using the term ‘type C variant’ in the classification field for this type (not prefixed with ‘Winchester’ as they don’t occur there).

It has an openwork sub-rectangular bit and often has a lozenge-shaped bow similar to that on other medieval padlock keys (both Winchester type A and type B). Its chief characteristic is an integrally cast T-shaped projection beyond the bit; the function of this is unknown. The stem often has a moulding on one face only, and again we don’t yet know whether this is functional, or merely decorative.

Post-medieval keys

Bows with a pair of internal projections are confined to the post-medieval period. Later post-medieval keys often had the bow shaped internally to a slight figure-of-eight form, as if they were intended for use with two fingers.

The Winchester type C key was adapted, perhaps in the 19th century, for use with the ‘French night latch’ (see explanation on the Queens’ College Cambridge website). We have a few of these recorded on the PAS database (see LANCUM-9C5462 illustrated above).