Forensics Expert from AZ Audit: Allegheny Co, Pennsylvania Had at Least 10,000 Anomalous Ballots

Forensic chemist and forensic document analyst expert Erich Speckin was one of the unknowns working behind the scenes during the Maricopa County Audit to analyze the paper and the ballot files that were used in the 4th largest county in America.  During this process, he revealed that at least 36,000 ballots were questionable and should be further examined.  Most of these ballots  (~25,000) were not printed from the official pdf, which caused anomalous discrepancies on these ballots to make them easily distinguishable from their authentic counterparts.

According to Speckin, the ballots printed from the original pdf would be “perfect”.  They would have perfectly straight lines with no breaks, etc.  However, when he examined the early and election day ballots, there were approximately 25,000 that had “breaks in all the same places” around the ovals.  And they were in the same place on all of the anomalous ballots.  You can see Speckin’s video about his work in Maricopa here (about 9 min).


Another anomaly that Speckin found in Maricopa Co dealt with batches that went almost entirely for one candidate.  Speckin found 61 batches each containing 200 ballots that were presented that had 0ver 90% of votes cast for a specific candidate.  In analyzing these batches, it was determined that 58 of the 61 had votes favoring Joe Biden by 90% or more and only 3 favoring Donald Trump by 90% or more.  This would mean that these combined batches went approximately 11,600 Biden to 600 Trump, in a state that has been red for approximately 40 years and in a county that was within 2.2% points of each other.

Fast Forward to Allegheny County, Pennsylvania

(video will play shortly after start despite “bad extension” warning)

On September 16, 2022, Erich Speckin released a report on Allegheny Co.’s ballots after doing a similar analysis.  Speckin notes in his report that he was only able to get access to the pdf images of the ballots, and not the physical ballots themselves, as seems to be the case almost everywhere when it comes to the 2020 election and the 2022 primaries.

“As a result of only having images, the forensic examination that can be performed is limited. It is always
more desirable to have the original evidence when possible. The examination conducted used the
available captured reliable evidence in the images to make the findings and arrive at the conclusions.”

When Speckin examined the mail-in ballot images, he found that a range of ballot IDs had a substantially inferior image quality compared to other ballot images provided.  Further, these inferior images all came from the same scanner, scanner 8520050452, according to the tabulation records provided.  Speckin writes:

“Many of these images have similar defects in the printing/copying process which would indicate common source of production, but different than the remaining examined ballots. Clearly in the mail in counting area, a random printing defect that may have existed in mailing out of ballots could not consistently come back and be counted on the same scanner. Therefore, a printing defect on the printing side of the ballot creation is not the likely cause of this anomaly.”

Speckin deduced that this could be caused by a defect in that particular scanner if it weren’t for a number of ballot images captured by the scanner that had no issues in the quality of the image or the repeatable defects.

The number of ballots that contain this visible defect is greater than 10,000, according to Speckin’s report.  Speckin, who has almost three decades of experience analyzing documents forensically, stated his professional opinion is that there needs to be a further review of the 700,000 ballots cast in Allegheny Co to determine what significance these findings have on the whole.  Speckin recommends examining the specific tabulator to determine the user of the tabulator, time frames of the scanning, and the cause of the poor images.

Speckin also recommends examining original ballots in order to look for impressions of other ballots pushing through as if they were filled out on top of each other, using infrared light to determine if more than one type of ink was used on each ballot (and see if its exclusive to a particular race), and to look at “color printed ballots to review the CPS code and yellow printed dots to determine similarities in the machine and the timing for the printing process.”

In Allegheny Co, most of the local state representative and senate races were within this margin of victory (<10,000 votes), however, not every voter in Allegheny votes for those races.  The race that would be interesting to look at would be the Conor Lamb/Sean Parnell race, where Lamb won by just over 10,000 votes with 430,000 votes cast.  Parnell claimed that, on election night, his staff determined it was mathematically impossible for Lamb to win and Lamb had allegedly called Parnell on election night to congratulate him on his victory.  But as the numbers began to trickle in for the days and weeks after the election, that changed.  Perhaps this should be further examined.  The PA Dept. of State sent out a memo last month requesting counties retain their 2020 election materials.


Photo of author
Follow me at, and @CannCon on TruthSocial

You can email Brian Lupo here, and read more of Brian Lupo's articles here.


Thanks for sharing!