I don't know about you, but I find it very hard to compute when the date is that marks the precise year before which we need to find prior art to knock these patents out. And when do the patents expire? So when he sent me an email, expressing some disappointment that a patent infringement lawsuit had been filed against Red Hat and Novell, I saw a chance to learn.

So I asked him those two questions, and according to him, all three expire in December next year, which makes it rather puzzling why they are using them in litigation. He also notes some fine points to add to our knowledge of how this crazy US patent regime works, like what a "terminal disclaimer" is. And I finally know what the probable date is for searching for prior art:

The short answer is that anything technically relevant and effectively dated before March 25, 1987 (Xerox's first application date which is a prima-facie invention date) is probably going to end up citable unless the inventors can antedate it, and anything before 25 March 1986 (the 'statutory bar' date) is definitely going to be citable.

There is, however, a long answer, and there are some fine points that you will find, I hope, as interesting as I did. The most important thing I learned was that prior art that is later than the starting date can in certain circumstances also be useful. Let me show you. It's complex, but hang in there until the end and it's worth it.

[ Update: News just in that the USPTO has rejected Amazon's 1-Click patent almost in its entirety. Amazon has a chance to respond, but it's an amazing story. Prior art was submitted by a guy in New Zealand with a request for reexamination, and it has knocked out most of the claims. 8 claims were knocked out by one article in Newsweek. So tally ho! Don't forget you can view all the patents by going here and looking for Exhibits A-C. ]

Here's the first email I got, in which he expressed sadness about the IP Innovations filing and then shared with me some things he thought would be helpful:

All three involved patents should expire December 10, 2008. This means that it's a bit of a puzzle why such a suit has been brought. Normal patentees would be on the whole unlikely to bring expensive patent infringement lawsuits on the basis of patents which are just about to expire.

That's mainly because the litigation can be expected to last until well after the expiry date. After patent expiry, the patentee can no longer get any injunction to stop further infringing trade, because that has by then become free to the public. The possibility of getting an injunction is normally one of the main incentives, often the biggest incentive, that motivates bringing suit for patent infringement. Once the patent has become history, the patentee is normally better advised to concentrate on other means of maintaining its competitive position.

After expiry, the only real issue left is usually compensation for infringements that took place before expiry. The expected back royalties or damages have to be very large before it can normally make business sense to sue over the historical usage.

I suppose that raises a question here whether this action is really being brought mainly for some ulterior reason, to try and make an example out of the defendants, or to cause them heavy legal expenses, or as some other kind of a demonstration, maybe an attempted softening-up to convince the defendants and the public that they are going to have continuing vulnerability to patent infringement problems for various parts of the GNU/Linux operating system.

I haven't looked at the claims or specifications here at all, nor which specific parts of the system have been targeted with the infringement claims, let alone any prior art. Novell and Red Hat are going to have to spend a lot of money having people do that.

You may possibly find the following informational items of some small use.

US 5072412 was issued December 10 1991. It is an old-law patent so it gets the possibility of a 17-year term counted from issue date, and should be due to expire December 10 2008.

The other two patents seem to be old-law patents too, because of their early (pre-June 1995) filing dates. (In principle they might possibly have lost old-law status due to certain possible events in the examination process, but the time between their filing and issue dates is short enough that such events did not likely happen. It would take inspection of the pre-issue prosecution history record at the USPTO to give the definitive answer.)

It's probably not worth checking the prosecution history just for the issue of old/new-law status, but it will become important for other issues of the scope of claim, and Novell-RedHat will certainly be looking at these details.

I think that the USPTO has not scanned the correspondence of patents as old as 5072412 and the others in this case, to put them on line, although this is happening with newer patent applications now.

The most of the pre-issue record that I can find now on-line for 5072412 and its friends is just a bare list of the dates of the correspondence items, which can be reached by first going to this page and then selecting 'public PAIR' and then the other suitable buttons and the patent number. What opens up is probably the limit of the information currently online, but it does include a semi-automated way of ordering hard copies from the USPTO.

About the expiry dates of the patents: December 10, 2008. The two additional patents depend on the continued existence of 5072412 because there are terminal disclaimers.

5533183 was granted 1996, but it has a terminal disclaimer that makes it expire on the same date as 5072412, i.e. December 10, 2008. 5394521 was granted 1995, and there is a terminal disclaimer there too, this time linked specifically to the date of Dec 10 2008 -- which comes to the same thing as in the other case.

Naturally, I found it encouraging that an injunction is unlikely in this case. That was the only part I was worrying about. I asked when is the killer date, the date we use as a marker to know what qualified as *prior* art, and how do you know when a patent expires? I had the answer on these patents and their expiry dates now because he'd told me, but I'd never be able to figure out how he reached that date or calculate any future patents, unless he explained more fully. So he did:

About patent dates, I'd better concentrate first on the expiry date question immediately relevant to the infringement suit against Novell and Red Hat. There are so many different kinds of dates in connection with patents -- with many of them subject to obscure qualifications and provisos -- that it would take several installments to cover even half of them. (By the way, I listened to the mp3 file on patent dates, referenced on Groklaw. I didn't get to see who the speaker was. While I would agree that the core of what he was saying is correct for the most common case, the format of his talk did mean that he left out provisos and qualifications which can be of dominating importance when -- maybe in the uncommon case -- they do apply. In searching for grounds of patent invalidity, one would certainly be interested in becoming aware of all of the detailed requirements, and searching astutely to see if any are not met!)

Patent expiry dates:

There's an old saying that computer people love standards -- that's why they have so many of them (and keep changing them!). In the same vein, you'd have to say that patent people (mostly the officials that run patent offices) just love date-systems; that's why they have so many of them (and they change, too). This little look is limited to date questions that determine patent expiry.

The law that governs the term of the patents that are in issue in the suit filed against Novell and Red Hat is a transitional provision contained in 35 USC 154(c)(1) (references are to sections of United States Code, Title 35, Patents, abbreviated 35 USC):

"The term of a patent that is in force on or that results from an application filed before the date that is 6 months after the date of the enactment of the Uruguay Round Agreements Act shall be the greater of the 20-year term as provided in subsection (a), or 17 years from grant, subject to any terminal disclaimers."

The effective date of the change referred to so indirectly in that subsection was in fact from and including 8 June 1995: 6 months after the cited act was signed by the President on 8 December 1994. (http://www.uspto.gov/web/offices/com/doc/uruguay/SUMMARY.html)

In spite of the fact that this is a already a long time ago, there still are plenty of patents around and alive to which the transitional law applies.

The 20-year term provision for patents granted on more recent applications is currently in 35 USC 154(a)(2):

"Subject to the payment of fees under this title, such grant shall be for a term beginning on the date on which the patent issues and ending 20 years from the date on which the application for the patent was filed in the United States or, if the application contains a specific reference to an earlier filed application or applications under section 120, 121, or 365(c) of this title, from the date on which the earliest such application was filed."

So how do these provisions apply to the patents in the suit here?

Let's start with patent number 5,072,412: The best version to look at is really the image of the printed granted patent (especially because, when there are important data about terminal disclaimers -- relevant to the two other patents involved in this suit, the information about them is often missing from the web text version of the patent). (For me, the image was here.)

The date of grant (issue) is given top right as Dec 10, 1991, just under the patent number, and next to index code (45). Following an old-fashioned practice, the label here is "Date of patent". (But this kind of label is perhaps nowadays more confusing than informative, when there are so many different kinds of dates associated with patents.)

The patent was granted on an application with its own number and date: these are found against index codes (21) and (22) further down the left hand column of the front page: No. 30,766, filed Mar 25, 1987.

In this particular patent, no earlier applications are mentioned: they occur in the case of the other patents, mentioned below.

From this information, we can conclude: that both of dates (21) and (45) are before the effective date mentioned in 35 USC 154(c)(1). So this patent gets a term which is the longer of: 20 years from the earliest application date mentioned, or 17 years from grant. Those two alternatives lead to March 25, 2007 and Dec. 10, 2008, so the expiry date will clearly be Dec 10, 2008.

Patent number 5,533,183 presents a more complicated picture: The issue date (45) is July 2, 1996. The application date under code (22) was Feb. 27, 1995. There is also a code (63) on the front page of the patent, containing mention of further patent applications, with dates of 1993, 1991, and 1987. These constitute a chain of patent applications, in which each is a continuation-parent of the next in date. This amounts to a claim of priority (under 35 USC 120), intended to give to the latest application in the chain, i.e. the one filed in February 1995, a benefit of priority, right back to the earliest application mentioned under code (63) -- which happens to be the same as the application on which the first patent in this little series was granted.

(This is not the place to discuss the interesting question of whether these claims of priority actually succeed legally, i.e. are justified by the substantive content of the earlier applications to which references are made. But the effects on the term of the patent are generated by just making the priority claims, irrespective, for this purpose, of whether those priority claims are justified in terms of substantive content or not.)

So, in summary, 5,533,183 was applied for before, but granted after, the effective date for change of patent term in June 1995. 35 USC 154(c)(1) therefore applies to this patent too. The ordinary process of applying that subsection would lead to a conclusion that the patent has the longer of 20 years from the earliest patent application date mentioned (25 March 2007) or 17 years from issue, which would look like July 2, 2013. But there is an extra factor operating here. The issue date is asterisked with reference to a notice that "The term of this patent shall not extend beyond the expiration date of Patent No. 5,072,412."

Terminal Disclaimers

This is called a terminal disclaimer: a disclaimer of every part of the patent right that would otherwise be operative after the date mentioned, and from then to the end of the patent term. The patent effectively expires earlier than the normal statutory date, and instead its life ends on the date specified by the terminal disclaimer. Why did the patent holder make a terminal disclaimer? The pre-grant correspondence will reveal the reason, but normally it is done to overcome a 'double-patenting' rejection. This is not the place to go into the legal basis of double-patenting rejections, or the reasons why the applicant filed claims that were not inventively distinct from each other, in two separate applications. But the effect of the disclaimer is that for practically every purpose (except the charging of official fees!) the two patents operate as if they were one, with a single expiry date and a single owner.

There could have been in principle yet another factor operating to affect the term of this patent. If in the period between June 1995 and the issue of the patent, the pre-grant process had included the filing of a 'continued prosecution (continuation) application' (CPA) or 'request for continued examination' (RCE), then the old-law status would have been lost, and the term would have been determined only by 35 USC 154(a)(2) -- subject, as always, to any terminal disclaimers. In that case, the patent would already have expired, on March 25, 2007.

The only way to find out if any such events happened between June 1995 and the grant of the patent is to look at the patent office record. A bare list of dates is to be found online. Go here, select the button for 'patent' and the patent number, and hit 'search'. Under 'transaction history' you see the dates and brief indication of the nature of the exchanges between examiner and applicant. There is nothing there in the nature of a CPA, RCE, or other kind of continuation application filed subsequently to June 1995. So it looks as if the old-law status was not lost.

The other patent 5394521: shows a history much like that of patent 5,533,183 in respect of its patent expiry position. There is a terminal disclaimer here too, it is drafted differently to refer directly to the date of Dec 10, 2008.

Being a simple soul, I then asked, so what is the killer date for prior art searching? And he elaborated:

'Killer' dates:

The short answer is that anything technically relevant and effectively dated before March 25, 1987 (Xerox's first application date which is a prima-facie invention date) is probably going to end up citable unless the inventors can antedate it, and anything before 25 March 1986 (the 'statutory bar' date) is definitely going to be citable.

See further details below.

The later two patents have continuation claims, for backdating, but one of them appears only to try to backdate to 1991. So the prima-facie and statutory bar dates there would be 9 Dec 1991 and 9 Dec 1990, unless the record proves that this was only a printing error by the patent office. Again, see below.

Also, if it turns out that there is anything wrong with the priority claims for either of the two later patents, then later prior art would become citable for whichever of those cases is affected.

I think you will able to see from the categories mentioned below, that it is possible to overlook really vital opportunities if one restricts one's attention only to events before the 'definitely citable' 'statutory bar' date, which is what is often done.

Normally, only the earlier of the two dates I have mentioned would be quoted, on the basis of making common assumptions. This masks potentially important facts, e.g. the assumptions may or may not be justified in the given case, and if they are not, then it likely opens up a real achilles heel on the validity questions.

Taking them patent by patent:

First patent 5072412: Earliest application date: 25 March 1987.

That means several things as regards potential prior art:

First, any art before 1987/3/25, the prima-facie invention date, which is the application date itself, is potentially a killer, insofar as the inventors fail to prove an earlier date of invention than that. Under the heading 35 USC 102(a), anything "known or used by others in this country (USA), or patented or described in a printed publication in this or a foreign country, before the invention thereof by the applicant for patent" is citable. Unless and until the inventors prove otherwise, the application date itself is treated as date of invention, so anything before _25 March 1987_ would be citable as prior art under this heading, unless and until the inventors prove that they antedated it.

Also citable as art in the same way (but with more probable difficulty in practice) would be any "invention made in this country (US) by another inventor who had not abandoned, suppressed, or concealed it" i.e. before 25 March 1987 or before the Xerox inventors' earlier date of invention if they can prove one. (Ref.: 35 USC 102(g)).

Some people take the view that inventors can always prove invention a whole year before they filed. Clearly this can't be generally true.

But if we take the pessimistic view, and suppose they can antedate by a whole year in this case: then, under the next heading, we look for anything that has been (35 USC102(b)) "patented or described in a printed publication in this or a foreign country or in public use or on sale in this country (US), more than one year prior to the date of the application for patent in the United States", i.e. anything in those categories before _25 March 1986_ would be citable as art, and then it doesn't matter what date of invention the inventors can or can't prove. That is why patent people often talk of a 'statutory bar', and a 'killer date'.

Note that 'printed publication' is a broader category than it initially looks to be. Its scope has been investigated by a number of appeal decisions, and it is certainly worth trying to bring up under this heading (for example) anything that can definitely be shown to have been distributed over the internet before a relevant date.

Additionally, if it turns out that certain kinds of patent documents contained relevant description, they would still be citable as art too, even if they were published later and/or missed by the patent office in its searches,:

• somebody else's earlier-filed US application for patent, i.e. something filed in the US before 25 March 1987 (or, if the [Xerox] applicants manage to prove an earlier date of invention than that, filed before whatever their invention date proves to be). As long as that other person's application gets as far as being officially published in the US under section 122(b) (USPTO online publications) or granted as a (possibly later) patent, even if it's not actually published until later, it is citable. (Reference 35 USC 102(e)).

• (but somebody else's US patent granted on an earlier- filed international application is excepted if it didn't reach the USPTO under 35 USC 371 before 25 March 1987 or earlier Xerox invention date if proved). (Reference 35 USC 102(e) as it used to be in force at the time relevant to these particular patents).

Second patent: 5,394,521: This was filed on May 6, 1993 (application 58,360) with a continuation claim to an application filed on Dec. 9, 1991 (application 805,343) and no other priority claim noted. That makes it look as if the prima-facie invention date was 9 Dec. 1991 and the statutory bar date 9 Dec. 1990. All that has been said above applies, but with relation to these two dates instead.

But, it does look as if we may possibly have here a printing error by the patent office, because the details on patent number 5,533,183 tell us that 805,343 was actually a continuation (division) of the 25 March 1987 application.

The details of the pre-grant correspondence on file under 5394521 at the patent office will show the true answer. If there has been a printing error and the priority claims should have been recorded as going further back, the same prima-facie invention date of March 25 1987 and statutory bar date of March 25, 1986 would apply as in the other cases.

If it turns out that any of the priority claims are not justified (because a relevant claim granted in the patent was inadequately described in one or more of the intremediate priority application documents), then the claim would only be awarded a later effective date, and later prior art would become citable against it.

Third patent 5,533,183: This was filed Feb. 27, 1995 (application 395,858) with a chain of priority claims with dates (58360) May 6, 1993, (805343) Dec. 9, 1991, and (30766) March 25, 1987.

So it looks as if the prima-facie invention date and the statutory bar date are the same as above, March 25, 1987 and March 25, 1986.

Again, if it turns out that any of the priority claims are not justified (because a relevant claim granted in the patent was inadequately described in one or more of the priority application documents) then the claim would only be awarded a later effective date, and later prior art would become citable against it.

As you see, it is way more complex than you and I thought. And even he sent a follow up message:

One reason for being fussy about not overlooking possible additional vulnerabilities of the later-granted patents in a series to later prior art than their nominal priority claims would indicate is the pattern of tactics that can lead to such later-granted patents arising in the first place.

If the applicant has had a fight with the examiner, then he will often take the grant of some claims, possibly narrow ones, as soon as the examiner offers - just so he ends up with at least something. But he develops worries that these claims may be so narrow that they can be easily designed around. So what he persists with, in his continuation applications, are likely to be claims that cover the ground that he is really worried about and which he fears was not effectively covered by the first patent to be granted. He does not worry overmuch that the examiner demands a terminal disclaimer, because his claims last the normal patent lifetime anyway (and for cases filed since 1995 terminal disclaimers often don't affect the lifetime at all).

But the claims in the later-granted patents can be affected by slightly more subtle causes of invalidity which might not have applied if the claims had been included in the first patent to issue. For example, the wording of the claim may have been 'developed' in a way that arguably is not supported by the initial application. In that case, it can possibly be proved to end up with a comparatively late date, in spite of the attempted priority claim, and be invalidated by late-dating prior art that can't touch the original application.

If the defendant's activities may be more vulnerable to the later-granted patents than to the earlier, it is specially important not to overlook art that is only arguably citable against the later-granted patents.

So, I am relying on you brainiacs to extrapolate from all this. And we have now a point in time to use for prior art searching in this case, subject to some further research on the second patent. But as he points out, if any of the priority claims are not justified, later prior art is useful too, so just provide anything at all that you know about. And if you have follow up questions, post them and I'll try to get answers for us. And if you want to review the prior art possibilities others have posted already, you'll find them here and here.