How old your furniture is, and how it was made
Several construction clues read against each other give a rough era. One clue on its own gives nothing, and the clue that disagrees with the rest is usually the most interesting thing on the piece.
- Start here: the essentials
- 1. Why one clue settles nothing
- 2. The finish is the best clue you have
- 3. What a drawer tells you
- 4. The other clues, and what this manual will not print
- 5. When the clues conflict
- 6. Labels, stamps and the thing that beats every clue
- 7. See it: Dana's fourth drawer
- 8. Practice: four clues and a range
- 9. Your piece
- 10. Check your work
- 11. Knowledge check and scenario
Start here Essentials
At minimum this chapter asks you to read three or four independent clues on one piece and write an era as a range rather than a year. The published-dates discussion and the labels detail are Go deeper on a first pass.
- Test the finish and read the answer as a date range. It is the clue with a real published source behind it (section 2, and Finding out what finish is on your furniture runs the test).
- Pull a drawer all the way out and look at four things. The joint, the tool marks, the bottom and the back (section 3).
- Look for the clue that disagrees with the others. It is usually worth more than the ones that agree (section 5).
- Write a range into the What it is sheet, with the clues beside it. Not a year (your condition report).
Minimum time: about 30 minutes. Done when: your report gives an era as a range, lists the clues behind it, and names anything that looks replaced.
1. Why one clue settles nothing Understand
This is the question everyone arrives with, and it is the chapter where this manual is least able to give you a chart. You should know why before you read any further, because the reason changes how you use everything below.
Dating furniture by construction is a real skill with a long tradition, and the clues are genuine: the way a board was sawn, the way a drawer was jointed, the kind of screw or nail used, the hardware, the finish. Each of these changed over time, so each carries information. The difficulty is not with the clues. It is with the dates attached to them.
When the research for this manual went looking for the dates, it found two things. First, no conservation body, museum laboratory, extension service or government publication that we could reach publishes a construction dating chart for furniture. The institutions that publish the material this manual leans on for everything else, which is to say the conservation bodies and the federal laboratories, do not publish this. Second, the trade and collector guides that do publish it disagree with each other by decades on the same clue. Not by a few years. By decades.
Convention So this manual prints no construction date ranges. Not because the clues are worthless, but because a date range is only as good as its source, and we could not find a source for these that we would be willing to put a reader's decision on. What the chapter teaches instead is the method, which is the part that transfers: read several independent clues, see whether they agree, and pay most attention to the one that does not.
That is less satisfying than a chart, and it is worth saying plainly that it is less satisfying. It is also what the evidence supports. A reader who wants the chart will find a dozen versions of it online within a minute; what this chapter can usefully add is the warning that those versions contradict each other, and the habit of treating any one of them as a rough indication rather than a measurement.
Nothing in this chapter dates a piece to a year, and nothing in it establishes authenticity or value. An era read from construction clues is a rough band with soft edges, and three clues agreeing is a stronger band than one clue alone, not a certainty. If the age of your piece matters for a decision about money, that is an appraiser's question and it comes before any work, not after. Who to hire for furniture work says which trade does what.
2. The finish is the best clue you have
The one clue in this chapter with a published source and a stated method behind it is the finish, and it is also the quickest to read. That makes it the place to start.
The method is a solvent test, published by Bob Flexner, a finishing writer, in Popular Woodworking in April 2016. Different finishes dissolve in different solvents, so putting a small amount of each on a hidden spot and watching what happens names the finish, or names the family it belongs to. Finding out what finish is on your furniture runs the test properly and the solvent sequence is the decision table; this section is about what the answer means for a date.
The periods Flexner gives, in his own article, are these. Before the nineteenth century makers used what was available, usually wax or linseed oil. By the 1820s shellac and other alcohol-soluble resins had become widely available, and shellac was the dominant furniture finish for about the next hundred years. Nitrocellulose lacquer arrived in the 1920s and slowly replaced it. By the 1960s further resins, mainly catalyzed finishes, began replacing lacquer, with polyester and UV-cured finishes later. Water-based finishes were introduced in the 1980s and 1990s to reduce solvent emissions.
| Test result | Finish | What it suggests |
|---|---|---|
| Softens in denatured alcohol | Shellac | Consistent with anything from the 1820s on, and most likely before about the 1920s if it is the original finish |
| Softens in lacquer thinner | Lacquer | Consistent with the 1920s on |
| Goes soft and tacky in lacquer thinner, toluene or xylene | Water-based finish | Consistent with the 1980s on |
| Nothing reacts | Varnish, or a newer solvent-resistant finish | A group rather than a date: oil-based varnish, polyurethane, catalyzed, polyester or UV-cured |
| Comes off in mineral spirits | Wax, over something else | Nothing about the date until the wax is off and the finish beneath is tested |
Two cautions, both of which come from the article itself rather than from this manual's caution. Test it The whole method rests on an assumption the author states openly: that you are dealing with an original finish rather than a refinished one. A refinished piece tests as its newest coat, so what you learn is the date of the last person who worked on it. And when nothing reacts, the honest reading is the one he gives, which is that the finish is varnish or one of the newer more-solvent-resistant finishes. That is a group spanning five coating families with a century between their ends, so it is not a date.
Physical principle One more point, easy to miss: the finish dates the finish, not the wood. A 1790 chest refinished in 1975 tests as whatever went on in 1975. This is not a flaw in the test. It is a second piece of information, and in some ways more useful than the first, because it tells you that somebody stripped the piece. A finish clue that disagrees with every other clue on the piece is probably telling you exactly that.
3. What a drawer tells you
A drawer is the richest single object on most pieces of case furniture, because it is small, it is made of several parts, most of it was left unfinished, and the whole of it was built to be taken out and looked at from every side. Pull one right out, set it on a table and look at four things.
The joint at the front corner
Look at how the side meets the front. On older case furniture this is usually a dovetail, a joint of interlocking wedge-shaped pins and tails. What to notice is not just that there are dovetails but what they look like: how many, whether they are evenly spaced or vary, whether the tails are narrow or wide, and whether there is a faint scribed line running across the end of the tails where the maker marked their depth.
Hand-cut and machine-cut dovetails look different, and the difference is visible without training. Machine-cut dovetails are regular: same size, same spacing, repeated. Hand-cut ones vary, often noticeably, and frequently carry the scribe line. Convention The trade reads hand-cut work as older and machine-cut as later, which is reasonable in direction. The dates attached to the transition are the ones that disagree across sources, so note which kind you have rather than what year it implies.
Other joints you may meet instead: a simple rabbet with nails, which is quick construction of any era; a machine-cut joint with rounded interlocking pins, which is a distinctive factory joint; and on newer work, dowels, biscuits, staples or knock-down fittings.
Tool marks on the unfinished surfaces
Look at the drawer side, the underside of the bottom and the inside of the back. Hand planing leaves slightly irregular, shallow, scalloped tracks. Machine planing leaves fine regular ripples. A circular saw leaves arcs. A straight reciprocating saw leaves straight parallel marks. Sanding leaves scratches that follow whatever the machine did.
These marks are real evidence about how the part was made. Again, this manual prints no dates for them, for the reason in section 1. Note the marks and photograph them.
The drawer bottom
Is it one solid board, several boards, plywood or hardboard? Which way does its grain run? Is it chamfered, that is, thinned at the edges so it can slide into a groove? A solid bottom in a groove is a construction that accommodates the wood's movement; a plywood or hardboard bottom nailed on is a different approach and a later one.
Whether the drawers match each other
This is the one most people skip and it is often the most informative. Pull every drawer, not one. Line them up. Are the joints the same type on all of them? Are the sides the same wood and thickness? Do the runners show the same wear? A set of drawers was made together, so differences between them mean something happened later.
4. The other clues, and what this manual will not print
The remaining clues are worth looking at and recording, and the manual is going to stop short of dating them. Here they are with what to observe.
| Clue | What to record | What it can indicate | What it does not settle |
|---|---|---|---|
| Saw marks on unfinished surfaces | Straight parallel marks, curved arcs, or none visible | The kind of saw used on that board | A date; the published ranges for saw types disagree across guides |
| Dovetails and other joints | Hand-cut or machine-cut, how many, scribe line present | Whether the part was made by hand or by machine | A date, and whether the whole piece matches that part |
| Screws | Slot or cross head, whether the threads are even, whether the tip is blunt or pointed, whether the slot is off-center | Rough manufacturing era, and often whether a screw is a later replacement | A date, and nothing whatever if the screw was replaced |
| Nails | Square or round in section, head shape | Broad manufacturing type | A date. This manual publishes no nail dates: see the note below |
| Hardware | Material, fixing method, whether holes match the current pulls, whether all pieces match | Whether hardware is original or replaced, which is often more useful than its age | The piece's date, since hardware is the most commonly replaced part of any piece |
| Glass and mirror plate | Thickness, whether the reflection is distorted, how it is held in | Whether the glass may be original to the piece | A date, and whether the glass and the frame are the same age |
| Wear patterns | Where the finish is worn through, where edges are rounded, whether wear is consistent with use | How the piece was used, and whether wear is plausible for the construction | Any date; wear is a function of use, not of years |
On nails specifically. There is a peer-reviewed literature in historical archaeology on dating sites from nails, and it is the right body of work for the question. This manual did not obtain it: the paper we wanted sits behind a publisher sign-in, and every route to it ended there. So we have not read it, we do not know what it concludes, and the chapter offers no nail dates rather than repeating the ones the collector guides give. If you need nail dating, that literature is where to look, through a library that can reach it.
Why the published ranges disagree, and what to do with that
When you go looking for the chart this chapter declines to print, you will find several, and they will not agree. The pattern worth understanding is why, because it tells you how much weight to give any of them.
Three reasons account for most of the disagreement. Invention is not adoption. A technology's first patent, its first commercial production and its ordinary use in everyday furniture can be separated by decades, and guides that quote an invention date as a dating threshold are quoting the earliest possible moment rather than the likely one. Adoption was uneven. A city workshop with machinery and a rural one working by hand were producing at the same time, so the same clue carries different dates in different places. Old stock and old habits persist. Materials sit in a shop for years and workers keep working the way they learned.
The practical upshot: treat any published range as a soft floor or a soft ceiling. A machine-made feature puts a floor under the part that has it, because the machine had to exist. A hand-made feature puts no ceiling on anything, because hand work continued alongside the machines.
Why the dates matter less than people think
It is worth asking what the date is for. In this manual, almost nothing downstream actually needs it. The material decides what can be done to the surface, which is What your piece of furniture is actually made of. The finish decides what can be cleaned and waxed, which is the finish chapter. The structure decides whether it can be used, which is Joints, racking and breaks. The hazards pass decides whether it comes indoors, which is What can hurt you in old furniture. The one decision that really turns on age is whether the piece might be worth something, and for that the answer is an appraiser rather than a chart, as What not to do to old furniture explains.
5. When the clues conflict
This is the part of the chapter that is actually worth your time, and it works even though the dates are unreliable, because it is about internal consistency rather than about absolute dates.
Read four or five clues. Most of them will sit together. One may not. The disagreeing clue is the finding.
What a disagreeing clue usually means, in rough order of likelihood:
- A repair or a replacement. One drawer rebuilt, a back replaced, new runners, a leg spliced. This is the commonest explanation and it is usually visible once you look for it: the replaced part is cleaner, differently worn, or jointed differently.
- A refinish. The finish clue disagrees with every construction clue. Somebody stripped and recoated the piece.
- Replaced hardware. Extremely common and usually obvious from filled or unused holes behind the current pulls.
- A marriage. Two pieces from different sources put together, most often a case and a top, or a bookcase and a base. Look at whether the two parts' woods, joinery and wear agree, and whether the join between them looks worked rather than made.
- A reproduction. A piece made later in an earlier style. Here the construction clues tend to agree with each other and disagree with the style, which is the opposite pattern from the ones above.
- Nothing. Sometimes a maker did something unusual, or used what was to hand, and there is no story.
Physical principle The one clue that resists this analysis is wear, because wear is honest in a way construction is not: it accumulates from use, and imitating it convincingly takes deliberate effort. A piece whose wear is in places that use would not reach, or whose wear is even where use would be uneven, is telling you something about how it was finished rather than how it was used. Note it; do not conclude from it.
Whatever you decide, write the conflict down rather than resolving it. "Three drawers hand-cut, one machine-cut" is a durable observation. "Fourth drawer is a replacement" is a conclusion, and it might be wrong.
6. Labels, stamps and the thing that beats every clue
Before spending an hour on construction clues, spend two minutes looking for writing. Makers' labels, stamps, stencils, brands, model numbers, patent dates, retailer tags, shipping marks and pencil writing turn up on the backs of cases, inside drawers, under seat rails, on the underside of tops and on the bottom of drawer bottoms.
A label settles more than every construction clue combined, and a "made in" mark can settle a question in the opposite direction from the one you hoped. Flexner's own article ends with the case worth remembering: a client's eighteenth-century Chippendale dining set, eight matching chairs, where no solvent touched the finish, not even acetone. Turning the table over found "Made in Korea" on the underside. The solvent test had told the truth, which was that the finish was modern; the underside told the rest.
A patent date is a particular case worth knowing: it is a floor, not a date of manufacture. A piece carrying a patent date was made at or after that date, possibly decades after, because manufacturers kept stamping patent numbers for as long as the patent stood and sometimes afterwards.
7. See it: Dana's fourth drawer See it
Dana's walnut chest. Inherited; nobody knows where it came from. Four drawers.
Known, from looking: denatured alcohol softens the finish on the underside of the top rail, so the finish is shellac. Three drawers have hand-cut dovetails with a visible scribe line; the fourth has machine-cut dovetails of even spacing. The fourth drawer's sides are a slightly different color from the other three. The back boards are solid, not plywood. The drawer bottoms are solid boards, chamfered at the edges. No label anywhere Dana could find.
Assumed, and they are Dana's assumptions: that the shellac is the original finish rather than a later recoat, which is the assumption the whole solvent method rests on and which Dana cannot test. And that the fourth drawer is a replacement rather than an original drawer rebuilt at some point.
Still open: the era. Under the assumption that the shellac is original, the finish is consistent with a piece made before about the 1920s and after the 1820s, which is a hundred-year band and the honest width of what Dana knows. The construction clues agree with "made by hand" without narrowing the band, because this manual prints no dates for them. Narrowing further is a job for someone who can handle the piece: a furniture conservator or an experienced dealer, and if money is in question, an appraiser.
What Dana writes on the report: "Era: shellac finish, hand-cut dovetails on three of four drawers, solid back boards. Consistent with pre-1920s if the finish is original, which is not established. Fourth drawer differs in joint and in secondary wood: replaced or rebuilt, not settled."
Why that is a good entry: it is a range, it lists its evidence, it names its load-bearing assumption, and it records the conflict without resolving it. In two years it will still be true.
Worth reading alongside this chapter:
- Using solvents to date furniture (Bob Flexner, Popular Woodworking, April 2016): the solvent method and the period dates in section 2, in the author's own words, including the Chippendale set with "Made in Korea" underneath. Note that the article gives no ventilation or flammability guidance; this manual supplies that from the conservation sources in What can hurt you in old furniture.
- Caring for Your Treasures (American Institute for Conservation): the furniture leaflet in this series is where this manual's conservation guidance comes from. Useful here for what it says about patterns of wear indicating the history of use, and what stripping and refinishing do to them.
8. Practice: four clues and a range Practice
Minimum: one piece, four clues, one range written down with its evidence. Time: about 40 minutes. Materials: a flashlight, a phone camera, a ruler, and your condition report open.
- Look for writing first. Two minutes, every hidden surface. If you find a label, photograph it and note what it says before doing anything else.
- Read the finish. Either from a test you have already run, or by going to Finding out what finish is on your furniture and running it now.
- Pull every drawer and line them up. Photograph one front corner joint and one side. Note joint type, tool marks, bottom construction, and whether all the drawers match.
- Record three more clues from the table in section 4, writing observations rather than dates: what the screws look like, whether the hardware holes match the hardware, where the wear is.
- Write one sentence with a range and one sentence with the conflict. If no clue conflicts, write that: a piece where everything agrees is also a finding.
The disagreement exercise. Look up one clue from section 4, say machine-cut dovetails, in three different published guides, and write down the date each gives. Then write the widest range the three of them cover together. This takes about fifteen minutes and is the most convincing thing in this chapter: the spread you find is the uncertainty you would have inherited by trusting any one of them. Keep the note with your report, because it is the reason your range is as wide as it is.
9. Your piece Your piece
10. Check your work Check your work
Criteria
- Is your answer a range rather than a year?
- Does the report list the clues the range rests on, so somebody else could disagree with your reading?
- Is the finish's assumption stated, namely that the finish is taken to be original unless something says otherwise?
- Does the report record any clue that disagrees with the others, as an observation rather than as a conclusion?
- Did you look for a label before spending time on construction?
Worked example
Dana's entry, checked against the criteria: it gives a band rather than a year, it names four clues, it states the original-finish assumption explicitly, and it records the fourth drawer's difference without deciding what caused it. The one criterion it meets weakly is the first: a hundred-year band is wide. That is the correct width given the evidence, and writing a narrower one would mean inventing precision.
Common mistakes
- Dating from style. Style was reproduced continuously, so a piece in an eighteenth-century style is evidence of a style, not of a century. The construction clues are independent of it, which is what makes them useful.
- Taking one clue as decisive. One clue is a hypothesis. Four clues that agree are a range.
- Resolving a conflict too early. Writing "replacement drawer" instead of "this drawer differs in two ways" loses the evidence and keeps only your guess.
- Treating a patent date as a date of manufacture. It is a floor.
- Dating the piece from the finish without flagging the assumption. The finish dates the finish.
What is still unresolved
This chapter cannot give you a year, and for most pieces nobody can without a label or documentation. It cannot tell you whether a piece is authentic, a reproduction or a marriage; it can only tell you which questions to ask about the internal consistency of what you are looking at. And it says nothing about value, which is What not to do to old furniture, and why for the argument and Who to hire for furniture work for who settles it.