POSNER, Circuit Judge.

The plaintiff (we'll call her the applicant) applied for Supplemental Security Income, which is a benefit for which low-income people who are aged, blind, or disabled are eligible. She was turned down by an administrative law judge of the Social Security Administration for the benefit she sought for years before she turned 55. But because of the less demanding showing of disability required of applicants that age and older, she was deemed to have become disabled when she reached 55. She appealed the partial denial unsuccessfully, first to the appeals council of the Social Security Administration and then to the district court, and she now appeals to us.

The applicant's treating physicians, together with three consultative physicians selected by the Social Security Administration who examined the applicant and studied her medical records, advised the administrative law judge that she suffers from fibromyalgia, spinal disk disease, "photo-phobia" (abnormal sensitivity to light), and other ailments unnecessary to discuss, and that as a result she walks haltingly, has difficulty gripping objects, experiences difficulty in rising from a sitting position, has trouble concentrating in a bright room or when looking at a computer screen, and as a result of this assemblage of impairments cannot do even light work on a full-time basis. If this is right she was disabled before she turned 55 and is therefore entitled to a back payment of Supplemental Security Income.

"Light work" is defined by the Social Security Administration as work that "involves lifting no more than 20 pounds at a time with frequent lifting or carrying of objects weighing up to 10 pounds. Even though the weight lifted may be very little, a job is in this category when it requires a good deal of walking or standing, or when it involves sitting most of the time with some pushing and pulling of arm or leg controls. To be considered capable of performing a full or wide range of light work, you must have the ability to do substantially all of these activities. If someone can do light work, we determine that he or she can also do sedentary work, unless there are additional limiting factors such as loss of fine dexterity or inability to sit for long periods of time." 20 C.F.R. § 404.1567. This is a pretty precise description of the type of work that, according to the findings by the doctors regarding the applicant's physical limitations, she can't perform.

But the administrative law judge brushed aside the physicians' findings. Typical was his statement that the opinion of Dr. Dauscher, one of the applicant's treating physicians, would be "given no significant weight, because the functional limitations are not supported by Dr. Dauscher's sparse treatment statement notes or by examination findings made by other physicians." The administrative law judge seems to have thought that a physician's evidence can be disregarded unless he has detailed notes to back it up and other physicians provide identical evidence even if they don't contradict him — in other words no credibility without corroboration. These are insufficient grounds for disbelieving the evidence of a qualified professional.

The administrative law judge discussed at greatest length the evidence of Dr. Michael Holton, one of the consultative physicians, saying that Holton had

This is garbled. Consider first the criticisms by Dr. Sands. Sands could not have been talking about Dr. Holton, because Sands's report preceded Holton's. The government's lawyer admitted this at the oral argument but speculated that it was a "scrivener's error" — that the administrative law judge had meant Dr. Ksionski when he said Dr. Holton. This is possible, but we can't assume it to be true on the basis of the lawyer's speculation.

Consider next the statement attributed by the administrative law judge to Holton that "manipulative abilities were normal." In fact Holton noted "grip strength" measurements of 31 pounds for the applicant's right hand and 11 pounds for her left, which are well below the normal range for women of the applicant's age. See, e.g., Virgil Mathiowetz et al., "Grip and Pinch Strength: Normative Data for Adults," 66 Archives of Physical Medicine and Rehabilitation 69, 71 (1985), www.fcesoftware. com/images/5_Grip_and_Pinch_Norms.pdf (visited Dec. 4, 2014). The applicant in our case may have areas of strength and be able to feel things ("normal sensation") without having the grip strength that she'd need at work. The administrative law judge failed to compare X-ray and MRI evidence presented by Holton that revealed spinal disease of sufficient gravity to engender the limitations on gait, gripping (an important manipulative ability), and rising (for example, getting up from a chair one is sitting in or straightening up after lifting something) that he found she had.

It's true that Holton reported that the applicant's "fine finger manipulative abilities appear normal." "`Fingering' involves picking, pinching, or otherwise working primarily with the fingers. It is needed to perform most unskilled sedentary jobs and to perform certain skilled and semiskilled jobs at all levels of exertion." Social Security Ruling 85-15:2. But Holton had also opined that the applicant would have trouble "handling," a finding that is consistent with reduced grip strength (indeed, gripping is a form of handling) and is an essential manipulative activity in a great many jobs. The Social Security ruling that we've just been quoting from explains that "handling (seizing, holding, grasping, turning or otherwise working primarily with the whole hand or hands) are activities required in almost all jobs. Significant limitations of reaching or handling, therefore, may eliminate a large number of occupations a person could otherwise do."

The district court's statement that "the ALJ's evaluation of Dr. Holton's opinion may not be perfect" is a considerable understatement. Coupled with the administrative law judge's unreasoned brush off of the evidence offered by the other consulting physicians, his confused rejection of Dr. Holton's evidence should have persuaded the district judge to reverse the denial of relief to the applicant and remand the matter to the Social Security Administration.

There is more that is wrong with the administrative law judge's opinion. The more involves an issue we discussed in Browning v. Colvin, 766 F.3d 702, 708-12 (7th Cir.2014), concerning testimony by vocational experts regarding the number of jobs in the local, state, and national economy that an applicant for social security disability benefits is capable of performing. The Social Security Administration does not try to determine whether an applicant would have any real chance of landing a job, even if physically and mentally capable of performing the work required by it, but it does require a determination of whether work that the applicant is capable of doing "exists in significant numbers" in the economy. 42 U.S.C. § 423(d)(2)(A); 20 C.F.R. § 416.966(a). If the only jobs that the applicant is physically and mentally capable of doing no longer exist in the American economy (such as pin setter, phrenologist, leech collector, milkman, pony express rider, and daguerreotypist), the applicant is disabled from working, and likewise, as a realistic matter, if there is an insignificant number of such jobs.

Asked at oral argument, the government lawyers in both social security disability cases argued before us on October 28 confessed ignorance of the source and accuracy of such statistics, about which we had expressed profound doubt in the Browning case. We are not alone in harboring such doubts. See Brault v. Social Security Administration, 683 F.3d 443, 446-47 (2d Cir.2012) (per curiam); Guiton v. Colvin, 546 Fed.Appx. 137, 143-45 (4th Cir.2013) (concurring opinion); Jon C. Dubin, "Overcoming Gridlock: Campbell After a Quarter-Century and Bureaucratically Rational Gap-Filling in Mass Justice Adjudication in the Social Security Administration's Disability Programs," 62 Administrative L.Rev. 937, 964-71 (2010); Peter J. Lemoine, "Crisis of Confidence: The Inade quacies of Vocational Evidence Presented at Social Security Disability Hearings, Part II," Social Security Forum, Sept. 2012, p. 4.

The administrative law judge found that the applicant was capable of performing "a restricted range of light work." On the basis of that finding a vocational expert testified that the applicant "would be able to perform the requirements of representative unskilled light occupations such as: cashier, with 1,000 such jobs existing in the Fort Wayne region [where the applicant lives] and 10,000 such jobs existing in Indiana; shipping and receiving weigher, with 200 jobs in the region and 2,000 jobs in Indiana; and production inspector, with 500 jobs in the region and 5,000 jobs in Indiana" (citations omitted). For unexplained reasons he didn't estimate the number of jobs in these categories in the nation as a whole.

The only public source that the vocational expert cited for the numbers we've just quoted was the Dictionary of Occupational Titles (4th ed.1991) (the "DOT" as it is called). He did testify that he had also relied on his "knowledge of the industry" — "my past experience, knowledge of these positions, employers that do accommodate for individuals with varying degrees of limitations or impairments, hence personal experience and labor market surveys [] account for a portion of my testimony." But he didn't explain how impressions from unspecified past experience and "knowledge" could enable him to determine numbers of particular jobs. Nor did he reveal what surveys he had relied upon and what they had shown.

As for his reference to the DOT, not only is that an obsolete catalog of jobs (most of the entries in it date back to 1977) but it contains no statistics regarding the number of jobs in a given job category that exist in the local, state, or national economy. For the numbers, vocational experts normally rely on a journal called the Occupational Employment Quarterly, published by a company called U.S. Publishing, although the vocational expert in this case did not mention the journal. The source of the journal's statistics is census data, and the Census Bureau reports not the number of jobs in each job category in the DOT but instead the number of jobs in a broader job category that includes some of the DOT's narrower categories. The vocational expert divides the number of jobs in the broad category by the number of finer categories within the broad category, and the result is his estimate of the number of jobs in the finer category, that is, the number of jobs the administrative law judge believes the applicant for benefits is capable of performing.

So if the broad category contains 10,000 jobs, and there are 20 finer categories within it, one of which consists of the jobs the applicant can perform, the vocational expert would estimate, and the administrative law judge accept, that there were 500 jobs in that category. That would be an arbitrary estimate, since there would be no basis for thinking that all the finer categories include the same number of jobs — namely, in our example, 10,000 divided by 500. (For this understanding of how the vocational experts arrive at their conclusions, see, besides the sources cited earlier, David F. Traver, "Cross-Examination of Vocational Expert on U.S. Publishing Data," Attorney Education Center, www. jameseducationcenter.com/articles/cross-examination-public-data/ (visited Dec. 4, 2014), reporting an admission by a vocational expert that this is indeed how vocational experts arrive at their numbers.)

We do not know how the vocational expert in this case calculated the numbers to which he testified. Nothing in the record enables us to verify those numbers, which the administrative law judge accepted.

While we're trying to solve or at least identify puzzles, we'll take a crack at one more — why it is that the vocational expert is required to estimate the number of jobs in the applicant's locality and region, as well as in the nation as a whole, that the applicant for benefits can perform. For if there is a substantial number of such jobs in the nation, the applicant's claim fails, no matter how few there are in his locality or region. We are guessing that the reason for requiring estimates of local and regional job availability as well as national is that the number of jobs of a particular type in the nation as a whole might be very small, yet if they were concentrated in the applicant's area he might have a significant opportunity for obtaining work that he was capable of performing even though people living elsewhere would not have that opportunity.

REVERSED AND REMANDED.