Without Undue Experimentation vs Without Any Experiments

by Dennis Crouch

I was rereading the Supreme Court’s recent enablement decision of Amgen Inc. v. Sanofi, 598 U.S. 594 (2023) and was struck by the Supreme Court’s statement that its 19th Century decision of Wood v. Underhill, 46 U.S. 1 (1847) “establish[ed] that a specification may call for a reasonable amount of experimentation to make and use a patented invention.”  This statement from Amgen is surprising because Chief Justice Taney’s decision in Wood includes a seemingly contrary statement that bars any experimentation:

The specification must be in such full, clear, and exact terms as to enable any one skilled in the art to which it appertains to compound and use the invention; that is to say, to compound and use it without making any experiments of his own.

Wood, 46 U.S. at 4.  Although Wood does allow for some extrapolation from the specification disclosure to cover broader claims, it expressly holds that a patent is void if it requires additional experimentation to understand its scope.

The contrary statement in Justice Gorsuch’s Amgen opinion appears largely lifted from the United States amicus brief in the case, which argued “Wood and Minerals Separation establish that a patent does not violate Section 112(a)’s enablement requirement merely because its specification requires some degree of experimentation to account for the nature of the patented invention or the field in which it arises.”  Although the US brief acknowledged the “seemingly categorical nature of Wood‘s articulation of the governing law” it then argues that Wood permits experimentation. The brief appears to conclude that by saying “without making any experimentation,” the Supreme Court in Wood actually meant that “undue” or “unreasonable” experimentation is prohibited. To understand the apparent contradiction, I thought it would be helpful to dig into some of the details of the case.

Background of the Invention

Wood’s U.S. Patent No. 97, originally issued in 1829 and later reissued in 1836, was directed to a process of mixing pulverized anthracite coal dust with clay before molding the bricks or tiles. In his patent specification, Wood stated:

Take of common anthracite coal, unburnt, such quantity as will best suit the kind of clay to be made into brick or tile, and mix the same, when well pulverized, with the clay before [it] is moulded; that clay which requires the most burning will require the greatest proportion of coal-dust; the exact proportion, therefore, cannot be specified; but, in general, three fourths of a bushel of coal-dust to one thousand brick will be correct. Some clay may require one eighth more, and some not exceeding a half-bushel.

The basic idea here is to use pulverized coal as a sort of internal fuel that is mixed throughout the brick or tile material, allowing it to burn more efficiently and evenly during the firing process.  Wood claimed that this composition provided benefits of “saving of fuel, and the more general diffusion of heat through the kiln, by which the whole contents are more equally burned.”

The Lower Court Decision

In 1842, Wood sued William Underhill and Aschel Gerow for infringement.  This was a big case with Wood alleging that the defendants had made 100 million bricks and 100 million tiles using his patented process. Wood requested damages of $30,000, being triple his alleged actual damages of $10,000.  This is over $1 million in 2024 dollars based upon a 2% annual inflation rate.

At trial, the defendants argued that Wood’s specification was insufficient “because no certain proportion for the mixture is pointed out, but only that such quantity of coal must be taken as will best suit the kind of clay to be made into brick or tile.” They contended that there was “no fixed rule by which the manufacturer can make the mixture, but that must be ascertained by experiments upon the clay” and that the claiming clause was “too vague and uncertain to sustain a patent.”  The circuit court agreed with the defendants and instructed the jury that the specification was too vague and uncertain to support the patent. The jury found for the defendants, and Wood appealed to the Supreme Court.  At the lower court, the defendants also argued that Wood’s claimed invention was not new, having been known and used by numerous individuals in England, Wales, Canada and the United States prior to Wood’s alleged invention, but that issue was left moot by the decision that the specification was insufficient.

The Supreme Court’s Decision

In its decision, the Supreme Court reversed in a unanimous opinion authored by Chief Justice Taney. The key issue, as framed by the Court, was “whether his description of the relative proportions of coal-dust and clay, as given in his specification, is upon the face of it too vague and uncertain to support a patent.”

The Court acknowledged that if a specification “gives only the names of the substances which are to be mixed together, without stating any relative proportion” or if the proportions “were stated ambiguously and vaguely” then it would be proper for a court to declare the patent void. This would be the case when “it would be evident, on the face of the specification, that no one could use the invention without first ascertaining by experiment the exact proportion of the different ingredients required to produce the result intended to be obtained.”

However, the Court found that this “degree of vagueness and uncertainty” did not exist in Wood’s specification. The Court noted that Wood provided “a certain proportion as a general rule; that is, three fourths of a bushel of coal-dust to one thousand bricks.” While Wood mentioned that some clay may require slightly more or less coal dust, the Court regarded these as “exceptions to the rule he has stated; and as applicable to those cases only where the clay has some peculiarity, and differs in quality from that ordinarily employed in making bricks.”

The Court explained that “the general rule is given with entire exactness in its terms” and the variations mentioned seemed intended “to guard the brick-maker against mistakes, into which he might fall if his clay was more or less hard to burn than the kind ordinarily employed in the manufacture.” The Court acknowledged the possibility that clay qualities may differ so widely that the specification would be of no value and the invention could not be used without experimentation, but stated that this was not apparent from the face of the specification and was a question of fact for the jury to decide based on expert evidence. But the bottom line is that:

The Circuit Court therefore erred in instructing the jury that the specification was too vague and uncertain to support the patent,—and its judgment must be reversed.

The case was remanded for a new trial.  As suggested here, Wood v. Underhill highlights important distinctions between legal and factual issues in patent cases and the respective roles of the judge and jury.

One key issue was whether the vagueness and uncertainty of Wood’s specification rendered the patent invalid on its face. The Supreme Court treated this as a question of law for the judge to decide. The trial judge had ruled the patent invalid due to the lack of a precise proportion of coal dust, and the Supreme Court reviewed this ruling without deference.  However, the Supreme Court also noted that if a specification’s adequacy depends on underlying factual questions, such as whether a person skilled in the art could practice the invention without undue experimentation, then those questions should be put to the jury. The Court explained:

In patents for machines the sufficiency of the description must, in general, be a question of fact to be determined by the jury. And this must also be the case in compositions of matter, where any of the ingredients mentioned in the specification do not always possess exactly the same properties in the same degree.

So while the ultimate legal conclusion on a specification’s adequacy is for the judge, subsidiary factual issues about the state of the art and the predictability of the results may need to be resolved by the jury.  I have not been able to find what actually happened on remand in the case.

Reconciling Wood and Amgen

At first glance, Chief Justice Taney’s statement in Wood that the specification must enable a skilled artisan “to compound and use it without making any experiments of his own” seems directly at odds with Justice Gorsuch’s statement in Amgen that Wood allows for “a reasonable amount of experimentation.”  But I think that the better way to see these cases as both standing for the a fundamental principle – a patent specification must provide enough clarity that a skilled artisan can practice the claimed invention, but it need not spell out every last detail.

In his 1953 patent law treatise, John Paxton Norman provided a narrow treatment of Wood, concluding that it required disclosure at least some embodiments that can be created without experimentation.  John Paxton Norman, A Treatise on the Law and Practice Relating to Letters Patents for Inventions (1853). He cites Wood for the proposition that the invention is not patentable “if the qualities of the ingredients generally differ so widely that the rule given is of no value, or if the improvement cannot be used without ascertaining in every case by experiment the proportions to be employed.” Id.  This interpretation of Wood also pushes against the full scope enablement  and written description requirement that has become so popular.

30 thoughts on “Without Undue Experimentation vs Without Any Experiments

  1. 8

    The issue, at least with antibodies, has become whether one can predict if the antibody will work as desired.

    Amgen enabled how to make and how to use legions of antibodies. What it did not, and could not, do was enable predicting if the antibody produced will work as intended.

    One cannot do so in a biological system. Too many variables that cannot be controlled or predicted. One always has to test.

    1. 8.1

      Too many…?

      That sounds like an excuse to continue to treat the (so-called) “grown-up arts” in a non-deterministic manner.

      Stirring the cauldron of witches brew went out in the Dark Ages.

  2. 7

    I think we are arguing over minutiae. Enablement problems in a spec are pretty rare, but I dealt with it regularly as an examiner in business methods. The USPTO always allowed for some experimentation when applying the In Re Wands factors for testing enablement. Most examiners didn’t even know how to do an enablement rejection back in the day which might be why enablement problems are coming up in litigation now. Hopefully, the Patent Academy has changed that now so enablement problems will become an even rarer vestige of patent law. We can now focus on how many angels can dance on the head of a pin.

    1. 7.1

      Frankly, I doubt that the ‘Academy’ is doing anywhere near close enough of a job.

      I actually relish 112 rejections, as I have yet to not have the examiner withdraw their rejection upon my response (the prima facie case to be establish is much more difficult than your typical examiner realizes).

    2. 7.2

      How did the Patent Academy teach you to determine the level of ordinary skill in the art as part of the Wands Factors (and in a manner divorced from the cited prior art)?

  3. 6

    I think I see the problem…

    1. 6.1

      The file did not upload as a .jpg US Patent No. 00000097.

      1. 6.2.1

        What a patent app should look like. With a claim attached.


          In that day and age, perhaps.

          But perhaps not (seeing that litigation still ensued – even then).

  4. 5

    Thank you for a great teaching lesson on the enablement decision in Wood v. Underhill, 46 U.S. 1 (1847) [applied in Amgen Inc. v. Sanofi, 598 U.S. 594 (2023)]!
    It is a great example of the difference between quoting singele sentences from a decision out of context – disregarding the actual fact situation being decided. Here, for example, the single-ingredient-variation [% coal dust to clay] simplicity of the experimentation, and the guidelines and boundaries provided to the POSITA for that simple and limited experimentation. Nor does it require that experimentation to be on humans, as with variable drug dosages.

    1. 5.1

      P.S. In my view [I had even started writing an article on it], one of the worst examples of that kind of erronius intepretation of prior patent law decisions occured in Fed. Cir. decisions on 35 U.S.C. 112(a) “best mode” over the years. The harmful effect on litigation due in part from that inconsisency between earlier and later Fed. Cir. “best mode” decisons lead to the AIA legislation rendering it uninforceable as a litigation defense even though it is still in 112.

  5. 4

    Looking in from Europe, what I don’t see here is the distinction between ability to practice the “invention” revealed in the detailed description and the “invention” corresponding to the full scope of the claim.

    Is there a difference between the amount of experimentation one can reasonably expect of the skilled reader who is trying to practise what’s in the detailed description, and the amount that one can expect of a reader trying to put into effect an embodiment of the claimed concept that is well within, but out on the periphery of, the claim and significantly different from what the detailed description describes?

    In other words, what does the case law say about when the amount of experimentation needed to satisfy the requirement for enablement over the full scope of the claim rises to the level of “due” (or is no longer “undue”)?

    1. 4.1

      The case law on this IS there – it is what I have pointed out for many years now: the double-edged sword of Person Having Ordinary Skill In The Art.

      The judicial branch writing patent (profanity) law in making that legal fiction stronger in their efforts to turn 103 into a cudgel, necessarily make the ‘scope’ of requiring less within an application (your due/undue notion) to be opposite in favor of the patentee.

      As usual, the sticking point is the hypocrisy when those who despise patents want it both ways: a powerful legal fiction for 103, and a weak legal fiction for 112.

      1. 4.1.1

        And as I always point out, no one at the USPTO actually determines the level of ordinary skill, either explicitly or implicitly. Even the PTAB judges just go by what the parties set forth.


          I was mentoring a junior practitioner a while back and he had just obtained his patent bar registration.

          He was answering an Office Action 103 rejection, and in his “reverence” of the MPEP, he was not aware that a rejection that does not properly address the Graham Factors was lacking a prima facie establishment of a proper rejection, and was the easiest push-back against the examiner.

          Not just practitioners, but the Office itself needs to up its game against the MPEP being written in a substandard manner.

          As one can readily tell by the (typical) examiners that post on this blog, typically (and sometimes not at all) have a fingernail depth grasp on actual law (or their responsibility to examine under that law).


            ? What Graham Factors does an examiner need to address in a 103 rejection in the normal application file with none of record because the applicant has not filed any declaration in support thereof?


              Support for the choices of prior art and the examiner’s combination thereof (you know, to address the level of ordinary skill determination at point in this discussion).

              How long ago did you abandon your registration number?


              You seem confused as to who has the burden of establishing a prima facie case.

  6. 3

    If you patent an algorithm is writing a computer program to implement it “experimentation” or is the algorithm just a desired result?

    1. 3.1

      Are you conflating design choices of the computing arts (hardware, software, firmware)?

      1. 3.1.1

        Good point. Hardware implementation of algorithms used to be necessary for many algorithms but as stored-program computers became faster and faster, more and more algorithms could be done in software.

        An example is the box that Texas Instruments made in the 1980s to process speech to produce the data for its speech synthesis IC, the TMS5220, which used Linear Predictive Coding (LPC). LPC uses much less data storage than digitized speech.

        The box used several TRW MPY-16 bipolar multiplier/accumulator ICs. They cost several hundred dollars each in 1980s money. Later, TI replaced the box with one using one of TI’s new DSPs. (DSPs are fast and programmable and were already becoming faster and incredibly cheap.)

        The MPY-16 typically took 125ns to multiply two 16-bit numbers, produce a 32-bit result, and add it to its accumulator. It used a 64-pin DIP with parallel inputs and output. It was a huge power-hungry monster but even a mainframe would take at least several ms (maybe several hundred ms) to do the same thing.

        See: link to cpushack.com

        I suspect the MPY-16 made several important military and space projects possible. link to apps.dtic.mil

        In the 1980s (or so) the debate was whether you should be able to get a patent on anything done with a computer. If you couldn’t, it would mean that an algorithm implemented in hardware was patentable but if it was implemented on a computer, even a general purpose computer, it wouldn’t be.

        You can go the other way. You can take any computer program and implement it completely in hardware. It might be a huge amount of hardware but it could be done.

        There is also the issue of whether a micro-programmed machine is hardware, software, or both.

        BTW, firmware is software that is already loaded in non-volatile memory. Firmware used to mean masked ROMs, nowadays it is almost always flash memory which will eventually suffer from bit rot and fade away. Modern flash memory might take 100 years but it will happen. Then your furnace, air conditioner, TVs, microwave ovens, and everything thing else will stop working.


          Those things will “not work” long before that time period (we actively build in the notion of planned obsolescence).

          Regardless – the ‘wares’ are nothing but patent-equivalent design choice.

          Always has been (no matter how IBM fought that, back in the day).


          There is also the issue of whether a micro-programmed machine is hardware, software, or both.

          The point is – explicitly – in the patent sense, this is expressly NOT an issue.

          Leastwise, logically, it should not be one, and IS one only to the extent that anti-patent propaganda has been espoused.

  7. 2

    “And this must also be the case in compositions of matter”

    Is 112 interpreted differently based on the statutory category of the invention? I don’t know if the patent act of that time was a direct precursor to 101 or the judge was using this phrase in another sense.

    1. 2.1

      In Robinson on Patents (1890), he often expounds upon various substantive rules as applied to the four categories of invention separately. It seems like a plausible idea to me.


          Scheme, you say…

  8. 1

    The basic idea here is to use pulverized coal as a sort of internal fuel that is mixed throughout the brick or tile material, allowing it to burn more efficiently and evenly during the firing process. Wood claimed that this composition provided benefits of “saving of fuel, and the more general diffusion of heat through the kiln, by which the whole contents are more equally burned.”

    The Gist here is, well, obvious.

    Burn fuel in a kiln?

    “Just apply it.”

    from the frying pan of 112 to the fire of 101?

    1. 1.1

      further, that damm grifter, not providing an exacting single physical structure of a claim….

      Why, it seems that he wants to climb a rung on the ladder of abstraction.

Leave a Reply

Your email address will not be published. Required fields are marked *

You can click here to Subscribe without commenting

Add a picture