Election Day

• Election Night Reporting: 145,539
• Cast Vote Records: 145,539
• Ballot Images: 145,564

Recount

• Election Night Reporting: 145,544
• Cast Vote Records: 145,545
• Ballot Images: 145,545

Aberrations

• Double counted ballots in original count 50
• Ballots not appearing in recount: 21
• New ballots not appearing in original: 76

General Observations

In the original count, there was a notable incident of ballot doubling, affecting a total of 125 ballots. The ballots were found to be direct copies of other sets of ballots that had been scanned earlier, effectively doubling three full sets of ballots. Thankfully, a hard-working employee caught two of the doubled batches and properly removed them from the final vote count. Leaving only one batch of 50 double-scanned ballots.

Recount Analysis

In the recount, Cherokee County saw the inclusion of 77 new ballots included. Most of these new ballots are highly suspicious and discussed in the next section. The new ballots came in 2 batches and will be analyzed individually.

• Woodstock Precinct - 50 absentee ballots
• Rosecreek Precinct - 26 computer ballots

Matching Cast Vote Records vs. TIF Images

There were some very interesting changes made to the Cast Vote Record during the original count, which makes for a lot of initial confusion in the analysis. But thanks to having full sets of ballot images as well as Cast Vote Records, I think we can fully see what happened during the vote counting process.

In the original cast vote record, we see that a full batch of 25 double-counted ballots were in the ballot images. Someone has caught this abnormality and care was taken to properly remove these ballot images from the official cast vote record.

In the original cast vote record, we see another set of 50 double-counted ballots that were in the ballot images. Someone also caught this occurrence and properly removed these ballots from the official vote records as well.

On the flip side of this, we see that 50 ballot records in the cast vote record, but the ballot images are missing. I checked for duplication ballots, but instead were removed from the list of ballot images for some unknown reason. When I explored the folder, it was blank. It’s important to note that these 50 ballots should have been properly counted in the final totals. I cannot see any pattern on why these ballot images were removed.

Now when you consider that 75 ballots were removed from the cast vote record, and an additional 50 records were added back, the ballot count vs. the record counts match up perfectly with a difference of 25.

The Woodstock Precinct Recount Ballots

In Cherokee County, a significant issue arose that could be one of the clearest signs of ballot manipulation. During the original count, three batches of ballots were double scanned and counted, but the operator caught 2 of the batches and removed them from the official totals. Unfortunately, there was another batch of 50 records that was missed. These double-counted ballots were all from the Woodstock precinct.

In this double scanned batch, every ballot was for the Woodstock precinct, except for one lone vote ballot from the Arnold Mill precinct. During the recount, a new batch of ballot records emerged. These new 50 ballots mirrored the precinct pattern of the original batch, with every ballot again from Woodstock, and a single ballot from Arnold Mill. The odds of such a precise duplication occurring by sheer coincidence were astronomically low. This strongly suggests a deliberate action to inject ballots into the count.

Further analysis revealed a discrepancy in the voting patterns between the original and new batches. The original batch had a split of 23 to 27 in presidential votes, whereas the new batch showed a different split of 19 to 30. This variation indicates that the new set of ballots were not copies nor duplicates.

This new batch of 50 ballots appear below. All of these new ballots come in a single batch of 50 records and are sequential in order. In the screenshot, I took two samples of the data to not show bias in the data, and to focus on the single Arnold Mill ballot appearing at the end

The Rosecreek Senator Test Ballots

Echoing a pattern observed five other Counties, Cherokee County also had a set of senatorial test ballots. This batch comprised 20 consecutive ballots from the Rosecreek precinct, each exclusively marked for the second senatorial choice, with no votes for other candidates or races. The presence of these senatorial ballots, identical in nature to those found in three other counties, were clearly an attempt to pad the votes in the recount by adding test ballots to the total vote count. When you consider that there were 21 ballots in the original count, but are not in the recount, you must question why these ballots were added to the recount.

Also note that the senatorial choices were chosen twice to create additional ballots. This is the only occurrence of double-senatorial ballots that I've seen so far in Georgia.

The Roseceek Presidential Recount Ballots

After examining the Rosecreek precinct ballots and having recently finished reviewing the Paulding chapter, I've noticed some intriguing patterns in ballot additions. In Paulding, a new set of ballots were added, one for each presidential candidate, to increase the ballot count, without adding bias to the presidential vote. Similarly, in the Rosecreek data, I found an additional six new ballots included in the recount. These were placed near the senatorial test ballots from the same tabulator. Interestingly, some of these ballots followed right after the senatorial test ballots. These ballots had identical voting choices which were not found elsewhere in the vote record.

Additionally, I observed two additional pairs of sequential ballots for Joe Biden and Donald Trump in later batches. These appeared a few batches apart, but still relatively close together. The Trump ballots were sequential, while the Biden ballots, although not sequential, were from the same tabulator.

Are these test ballots added to the recount?

6 New Ballots created and added to the Recount

Comparison Against Official Numbers

In the analysis of Cherokee County's ballot count, a notable discrepancy surfaced between the official count and the cast vote records and images. The official results registered 333 fewer ballots than what the cast vote record indicated. Initially, I considered the possibility that these might be provisional ballots. However, this theory didn't hold, as there were only 26 provisional ballots recorded in the first machine count.

The recount ballots vs. the official results was considerably closer at only 79, but still no explanation on why it doesn't match the official released numbers.

A special thanks to Troy Horton for first identifying the added recount ballots and notifying the county. He has spent countless hours working with the Cherokee County Election Commission on identifying and resolving these issues.