27

Here is an image of a civil rights photographer named Cecil Williams drinking from a water fountain with a "WHITE ONLY" sign on it.

(c) Cecil Williams

The image is dated to 1956. The sign appears translucent and the words "WHITE ONLY" appear very professionally lettered. There are no fasteners connecting the sign to the fountain.

This has been posted to Reddit multiple times, and each time there is at least one comment suggesting that the sign is Photoshopped. For instance, from 10 years ago:

The sign looks shopped. Edges are very blurry.

[reply] The photo is a well documented historic photo by renowned civil rights photographer Cecil Williams. The native South Carolinian's body of work forms one of the nation’s most comprehensive photo collections on the civil rights era. He is regularly asked to speak at such universities as Clemson University and Columbia College on this and other famous photos he has taken in his storied career. It is absolutely not shopped.

This is apparently a famous photo (?) and the photographer has published it in a book and recreated it in a diorama in his museum.

museum

In 2024, a description of the photo appeared in the book Injustice in Focus: The Civil Rights Photography of Cecil Williams published by University of South Carolina Press. Williams describes the fountain as located at a service station on U.S. Route 21.

Williams explained that the station appeared deserted, and that his fresh experience as a civil rights photographer gave him the idea to get photographed taking a forbidden drink, which seems quite plausible to me. However, my skepticism that a cheap, unmanned service station had such an extremely professional, clean sign on its fountain inspires me to think that the sign may have been Photoshopped in recently to make the photographer's intent more clear. For comparison, here is a real photo of segregation signs on two drinking fountains.

Has the image been manipulated?

1
  • 2
    Comments have been moved to chat; please do not continue the discussion here. Before posting a comment below this one, please review the purposes of comments. Comments that do not request clarification or suggest improvements usually belong as an answer, on Skeptics Meta, or in Skeptics Chat. Comments continuing discussion may be removed.
    – Oddthinking
    Commented Feb 5 at 1:18

6 Answers 6

48

I haven't found a definitive way to date the photo, but the evidence put forward for it being fake is rather thin:

  • There doesn't seem to be any controversy that segregated drinking fountains existed. Getty Images has several photos showing different styles of sign, such as this one with a stencilled sign saying "For Colored Only" and this one showing several printed signs.
  • Nor does there seem to be any reason to doubt the date of the portrait, or the identity of the subject. It is easy to find photos by Cecil Williams than of him, but this one bears a striking resemblance.
  • There's no reason to assume the wording of such signs would be standardised, or that "White Only" would be an unnatural wording. See for example this photo of a laundromat dated as 1963.
  • The typeface is not particularly modern - it looks like Times New Roman, which was created in 1932.
  • The idea that a cheap gas station would not have a "professional" sign implies the owner would have to create it themselves, or pay high prices for printing. Printing and sign-making were mature industries by the 1950s, and it's quite possible that a "White Only" sign was something available "off the shelf", in the same way as a "Ladies Restroom" sign.

That leaves us with supposed image artefacts. Searching for details about the image I found a different version of the image available from Getty Images. Here are cropped and zoomed details of the sign from the two versions:

Crop from version posted on Reddit and included in the question:

Detail of sign on Reddit image in question

Crop from preview available on Getty Images:

Detail of sign on image preview in Getty Images

Some things to note:

  • The position of the sign, and the text, are identical in the two versions.
  • The Getty version shows what appear to be four pins holding the sign to the fountain, removing the supposed mystery of how the sign is attached.
  • Whereas the Getty version shows the graininess of a chemical film, and generally darker shading, the Reddit version is more uniformly blurry, as would be expected from a low-resolution digitisation.
  • The exception is the text, which on the Reddit image is sharp and crisply black; in contrast, the Getty version has imperfections, mostly white spots on the black text.
  • The contrast is stronger on the Getty version, with the "fuzzy edge" around the sign more clearly including a shadow at the bottom and right, consistent with the lighting in the rest of the image.

As a final comparison, here's a different area of the image, which nobody would have a reason to deliberately fake; Reddit version on the left, Getty on the right:

Detail of dangerous socket on Reddit image Detail of dangerous socket on Getty image

The Getty image clearly has higher detail in the bricks, but simultaneously more contrast between the black plug and the light-coloured socket.

This is all entirely consistent with the Getty image being a high-resolution scan of an original print, which included the sign and its pins. Whereas the Reddit image is a lower-quality scan, or a low-resolution copy of the same scan, where the sign has been digitally "cleaned up" to make it more legible. (That makes the answer to the question in the title technically "yes", the version on Reddit was digitally manipulated; but the answer to the implied question of whether the sign is fake is "no".)

7
3

In the comments to my other answer, someone suggested reaching out to Thomas Phinney, an expert employed in legal cases to identify fonts and date documents through typography (thefontdetective.com). I emailed him, and he replied with a lengthy analysis. The user Copilot asked to see Phinney's email, so I asked him for permission, and in reply he revised and extended his analysis to read as follows:

On the fonts/typography side, it sure looks like Times Bold. That is not out of period at all (it first came out back in 1932), but it is mildly surprising to me that a 1956 sign at a gas station would have professional printing at such a large size. I would estimate that is about 180 pt type. Hand-set metal typefaces generally topped out at 108 pt or less. Wood type could get up to that size or much larger, though such subtle and thin serifs were uncommon in wood type fonts. 1956 would be pretty early for phototypesetting in that kind of usage, and it was not normally so large, either.

Most such signs were either hand-painted by staff, or by a professional sign-painter. The signs linked from your example seem like non-pro work, a bit uneven. Even less than wood type, hand-painted signs will never look exactly like a standard typeface. The translucency also seems unusual. Why can we see the metal side of the fountain and its discolorations through the plaque?

HOWEVER, I note also that I see another version of the photo online with some key differences. It has marks like fasteners near the corners. The white is not as translucent. I certainly wonder why there are two versions of the photo out there. (I understand about the colorized version, that is another issue.)

https://www.facebook.com/photo?fbid=2226685900770598&set=pcb.2226686147437240 https://www.facebook.com/photo/?fbid=2226685924103929&set=pcb.2226686147437240

Even in those images, it is odd how blurry the edges of the sign are, when so much else is crisp and perfectly in focus.

I have no overall opinion, because there are aspects of this that I would need to research more before I had a strong take on whether the sign might seem inauthentic. I do see a few things that seem odd enough that one would be well advised to consult an expert on signs.

0
1

The version in "Unforgettable: Life Hope Bravery" (on the Neema Fine Art Gallery Facebook page) shows more detail in the watch face (probably glare from either flash or sun is the reason for the whiteout) and less crisp lines on the sink itself. I think that we can conclude that the photo itself is original, but your version has been modified – possibly resharpened as one commentator mentioned.

Regarding the typeface: How were protest signs in the 30s-50s U.S. so beautifully designed and lettered? on Reddit explains why protest signs in the sixties looked so professional. Essentially professional signmakers were much more prevalent and in particular a service station would have had access to corporate suppliers.

8
  • 2
    I don't see any detail visible in the watch face in that image.
    – IMSoP
    Commented Feb 5 at 21:42
  • To pile confusion on confusion, the lettering on THAT image (the one in the art book) also looks foolishly sharpened to me!
    – Fattie
    Commented Feb 5 at 23:19
  • You need to be logged in on Facebook to your first link. That's not ideal for this site which assumes that answers are (a) self-dependent and (b) as persistent against link rot as possible.
    – Schmuddi
    Commented Feb 7 at 7:57
  • @Schmuddi I added the relevant details from the Facebook page as text, which is as far as I can go to prevent link rot without copying the photo. If you have another link which is not Facebook I am willing to add that if you prefer.
    – AlDante
    Commented Feb 7 at 16:28
  • @AlDante: Well, since I haven't logged in on Facebook to see the photo, I'm simply in no position to provide a better link as I don't know what "this version" refers to. Couldn't you try to find a better link that doesn't require a login?
    – Schmuddi
    Commented Feb 7 at 21:02
1

Yes, it was shopped

The text and 'paper' were added. The reason the sign was described by some as 'translucent' is that it has the same kind of structure as the rest of the stand, so it appears like a milky overlay. Yet this is due to the 'sign' being shopped from a part of the stand, copied and then lightened, and moved a few cm up-and-left. The text and 'pins' were added in a separate step.

Proof:

As you can see on the pic i inserted below, the perspective lines of the exposed brick&mortar wall and the window ('architecture', blue lines) and of the ceramic part of the fountain ('ceramic', lilac lines) coincide ok (not perfect, but close; the ceramics part is probably a bit rotated around the vertical axis with respect to the architecture, which is possible as it seems not directly affixed to the wall itself).

The perspective lines of the sign ('paper', green lines) are way off (which could be explained by the sign-side of the stand not being a plane parallel to the wall, for instance rotated slightly (like a door that is slightly ajar), though, for the vanishing point to creep away from the image, it would have to be more parallel to the observer which is not borne out by the depth of the shadow under the sink), and the perspective of the text is completely off, from both the paper and the architecture.

If the text was on the paper and thus the paper and the text were on the same plane they would have the the same vanishing points. They do not. (The pins, for some reason, perspectively fit well to the architecture, but neither to paper nor text)

I added dots in the color of the perspective lines to signify the vanishing points of the respective lines, the architecture lines are only loosely meeting in the same area, thus the bigger dot. Note that the individual perspective-lines of each text line (upper and lower border of one line) meet far out (red dot in first image), but the perspective lines of the two words never meet, they actually diverge slightly...

Observe the blue architecture lines meeting close to the image, the ceramics slightly further out, and text and paper respectively far off from themselves and everything else: Overview, with dots for vanishing points Total, for orientation

As a reference, this is how it looks when the text on a sign is actually printed on a physical paper, and that paper is affixed to a rectangular-ish object. Note that the paper is actually a bit creased, and the text more blurry than in the original photo discussed above, still, perspective lines vanish at roughly the same distance, especially red&green (text&paper): enter image description here

Detail to show how i fitted the lines to the text and paper; observe how the lilac line (top) and red, yellow and green lines do not fit into the depicted scene at their respective angles: Detail of Text, to see how lines align to characters

Now the structure : here a version with the sign in low brightness & high contrast (ignore perspective-lines in this): Detail with sign in high contrast, showing similarity of patterns Now observe (in the above picture) the little bright 'L' shaped structure on the lower right of the lower right pin, that has a twin on the outside of the sign, lower and to the left; I fit a pink line in there that connects those clones. The same displacement is true for other, less recognizable bright spots (same pink marker used by me).

I cut the sign, and shifted it so those details come to overlap. Now observe that various vertical structures between sign and outside the sign suddenly line up - so this was where the sign was lifted from.(ignore lines in this) Detail with shifted sign, showing probable copy origin

15
  • 1
    Thanks for the had work, but I personally don't see how this is conclusive. The lines seem to be in the margin of error around the focus point. You should try tracing the lines that shoot out of the bricks being overlapped by the red lines you've drawn framing the text. Commented Mar 4 at 18:40
  • 1
    In your last photo, you appear to have 'eaten' more pixels into the 'O' than into the 'Y' in placing your red lines, especially the bottom one. That can move the intersection point significantly IMHO, given the small angle involved. But, TBH, the opposite seems to have happened with the 'W' and the 'E', so ???? Commented Mar 4 at 18:49
  • 4
    Is your analysis of perspective lines an established, reliable way of identifying digital forgeries? I'm absolutely no expert in this field, but I can think of a number of factors that might impact such an analysis (e.g. lens distortion, uncertainty about the physical properties of the objects shown, human variance in deciding on the slopes – for reference, the lines that I just drew in Inkscape don't always align nicely with your lines). Do you have a reference that discusses your analytic approach?
    – Schmuddi
    Commented Mar 6 at 11:15
  • 1
    Interesting, it's not just the L shaped white spot: there are at least 2 other white spots that are duplicated at the same offset.
    – Dan Getz
    Commented Mar 6 at 12:08
  • 1
    @Schmuddi: this older paper farid.berkeley.edu/downloads/publications/tr06.pdf uses it on objects with assumed/known geometry like rectangles (see e.g. fig 6) That's what we're trying to do here, but yeah, depends on geometry of the underlying objects being straight enough. The appendix to that paper also discusses geometric distortion due to the lens. I'd venture a guess that's a non-trivial factor for a medium-format camera like what was likely used here (based on Cecil's typical work camera.) Commented Mar 7 at 17:47
-1

This sign could not have been produced with typesetting in the 1950s, not even hot metal typesetting:

When printing large scale letters in posters etc. the metal type would have proved too heavy and economically unviable. Thus, large scale type was made as carved wood blocks as well as ceramics plates. (Wikipedia)

Even if this sign was mass produced(!) through photostat or photo-lettering processes, the original would have been hand-drawn:

Photo-lettering is a mechanical process, useless in itself unless supervised by persons who thoroughly understand designing with letters. (Alexander Nesbitt, The History and Technique of Lettering, Dover Publications, 1957, p.189)

Lettering artists would have reproduced the typeface from model books, and spaced the letters themselves, according to their whim:

If the artist considers the entire area of white space between letters he will soon acquire "feeling" in spacing. (Albert Cavanagh, Lettering and Alphabets, Dover Publications, 1946, p.9)

[In instances of distortions in spacing] the trained designer will make an almost intuitive adjustment... (Alexander Nesbitt, The History and Technique of Lettering, Dover Publications, 1957, p.228)

There were various tools to shrink or enlarge the typeface, but not to automate the spacing. In my opinion it's extremely unlikely that the spacing in hand lettering could be reproduced with the default settings on a computer.

As a demonstration of this, here are the default MacOS settings for Times New Roman overlaid on the manual spacing appearing in a manually set type specimen from the 1930s or 40s. I've used the Transform tool so that the first and last letters on each line are perfect fits.

enter image description here


With that in mind, I downloaded The Gimp, which I have not used in about 10 years, and tried to use the Text tool to imitate the shape of the text.

enter image description here

On my first attempt, I got a pretty good match for the W. I then used the Perspective tool to try to map it to the sign, and started working on the second line. I actually had more trouble with the Gimp's weird new GUI than I did copying the image.

enter image description here

As an inexperienced Gimp user, I was able to duplicate the lettering in about five minutes. The font was kept at default settings. The only tools I used were the Text tool and the Perspective tool.

enter image description here

In my opinion it is extremely unlikely that hand-lettering could be precisely reproduced by the default MacOS settings of computer typography.

33
  • 9
    I am not sure what conclusion you want us to draw from this. The argument seems to be similar to "This can't be an original Picasso, because I am a skilled forger and I just made replica of it that looks very similar."
    – Oddthinking
    Commented Feb 8 at 4:23
  • 2
    A "2D perspective tool" seems like an oxymoron: the tool is used to imitate the look of a three-dimensional object. It's exactly the tool you'd expect to need to reproduce a flat object photographed at a simple angle (or to straighten such an image, as I've done before to take a copy of a document when I didn't have a scanner available).
    – IMSoP
    Commented Feb 8 at 7:48
  • 2
    Why not? It's black Times New Roman text on a white background, what do you expect to be different about a 1950s version of it? A $1 bill is designed to be hard to recreate; a cheap sign is not.
    – IMSoP
    Commented Feb 8 at 10:01
  • 3
    @Avery If you can find a good source for "moveable type was generally not available in the size shown", and some other reasons why it would be difficult for the sign to be genuine, that would be a much stronger answer than one where you show that it would be easy to fake. In other words, you need to refute the null hypothesis, which is "this is a 1950s photo of a sign on a fountain"; and you need to do so with references, not experiments.
    – IMSoP
    Commented Feb 8 at 10:55
  • 2
    Actually, I've asked graphicdesign.stackexchange.com/questions/163440/… Hopefully they don't close it. Commented Feb 8 at 15:59
-10

Famously, fonts can be used to help date a document.

https://news.sophos.com/en-us/2019/01/17/telltale-font-scuppers-bankruptcy-trust-claim/

It's not conclusive since this boils down to eyeballing the font but it seems to match the font EucrosiaUPC that lists a 1992 copyright licensed to microsoft.

enter image description here

enter image description here

edit: I think IMSoP in the comments is correct. I suspect it matched to italic EucrosiaUPC due to the angle of the letters.

It's most likely just bold times new roman.

17
  • 7
    This would be more convincinf if you could show some distinguishing feature of the text that makes it match this font more than, for instance Times New Roman (created in 1932): i.sstatic.net/w04Ei.png
    – IMSoP
    Commented Feb 6 at 15:56
  • 21
    In fact, 2 minutes of research reveals that EucrosiaUPC is a font containing only Thai characters, and evidently contains Times New Roman fallback characters for Latin text (if you have a copy of Microsoft Word to hand, you can play with it for yourself). Why whatever uncredited tool you used picked such an obscure type face, I've no idea. Maybe the noise and perspective distortion in the image threw it off in some way.
    – IMSoP
    Commented Feb 6 at 16:10
  • 6
    I think it's most likely it matched multiple fonts and EucrosiaUPC was first alphabetically
    – CJR
    Commented Feb 6 at 16:32
  • 4
    @CJR Or EucrosiaUPC may be the only one that contains "italic" versions of the characters, created by slanting the original ones. Of course, in the photo, the characters are only slanted because the fountain was not photographed head-on, so the characters themselves are not actually italic.
    – Laurel
    Commented Feb 6 at 17:08
  • 6
    @Fattie I opened MS Word, typed "WHITE ONLY" twice, selected Times New Roman for one, EucrosiaUPC for the other. Other than a slightly different letter height, they are completely identical; it's pretty clear that Times New Roman (a pre-computer font that is presumably no longer under copyright) is used for the Latin characters. Does it really seem more likely to you that someone trying to fake a 1950s sign used an obscure font which Microsoft provides for its Thai support?
    – IMSoP
    Commented Feb 6 at 18:22

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .