Suppose that in an experimental study an inventor has found that an element x is critical to a new discovery. The inventor has discovered good results when x has a numerical value of, say, 1. Based on these good results, the inventor submits an invention disclosure to his company’s patent department.
At a later time, a patent practitioner starts the process of converting the invention disclosure into a patent application. In an effort to broaden the scope of the claims, the patent practitioner asks the inventor for a range of values for element x. At this point, the inventor has moved on to other projects. The inventor had conducted the experimental study with x=1 and stopped when good results were achieved. The inventor thinks x could be in a range from 0.5 to 7.5 but does not have any hard data to support this range. Nonetheless, the inventor delivers the prophetic range to the patent practitioner, who faithfully renders the range in the specification with the customary preferable range, more preferable range, and most preferable range verbiage.
Later on, after receiving a rejection from the patent office, the patent practitioner realizes that a range of values of x in the claims might help distinguish the claimed invention over the prior art. The patent practitioner proceeds to amend the claims to include the range of values of x. The range was specified verbatim in the specification—there should be no trouble with this approach right?
BAM! The examiner issues a rejection under 35 USC 112, stating that the claims are not supported by the specification because the disclosed examples show the claimed results to be true for only x=1. The examiner might have made other rejections, but 112 rejection is enough to worry about for now.
The patent practitioner might not like the 112 rejection, but who can really say that the examiner is off base? Isn’t there some rule that if a researcher wishes to make some conclusion based on an experimental study, the researcher should have adequate experimental data to support the conclusion? If the inventor wishes to claim that his invention has desirable results when an element x of the invention has values between 0.5 and 7.5, then the inventor must be prepared to show that this is true.
One way of showing this would be to demonstrate that at least the results are true at the endpoints of the range and, if possible, also at the midpoint of the range. Of course, if the solution space is highly nonlinear or discontinuous, this may not even be sufficient. However, it is clear that if the only data available are for when x=1, then the examiner can express doubt that the results are actually true for other values of x, and it will be up to the inventor to prove the examiner wrong.
The lesson to be learnt is that inventors should design experiments with a view for patenting, or even as a matter of good scientific practice. While in the real world it may be sufficient to just find a value that works, in the patenting world more is required if broad patent coverage is being sought.
As a matter of good scientific and patenting practices, if a phenomenon has been observed at a specific parameter value, the inventor should check whether the phenomenon also occurs at another parameter value before making any generalizations. The inventor must go beyond finding that good or optimum value that has the desired result, to finding a range of good or optimum values that have the desired result. The inventor might also want to generate hard data for values outside of the good or optimum range that do not work well, should it be necessary to prove non-obviousness at a later time.