Title	Description
ID	Comprises of Poll station ID Some kind of device ID Date stamp Time stamp We would have expected these files to be unique for each poll station, but noticed some poll stations that have a multiple of these files It looked like the PDF image was the same, so it could be that form 34A was resent by the KIEMS kit? It means that the KIEMS KIT stamps the date when it sends the form, or, The server that receives the Form34A from KIEMS KIT stamps it based on received time The ID points to the PDF on IEBC servers.. Please check to see whether your poll stations form34 is original
Station	The name of the poll station. Some poll stations are located at the same place, so will have a 1,2,3.. appended to the name. This data was lifted from a document that did not have the numbers, so we haven't appended them. Stations in the same precinct should have similar turnout & number registered voters
Last Modified	The time that the form34A was last modified. We don't understand why the PDF that was signed at the KIEMS kit has a different modification time from the time it was signed At times, this 'modification time' is earlier than the time on the filename - by almost 2 minutes This is how we can tell that the modification was done on a device not part of the IEBC infrastructure
Delay	The time between that on the form34A filename, and when the file was last modified. We don't understand why the PDF that was signed at the KIEMS kit has a different modification time from the time it was signed This is a negative value for about 6,000 forms The negative value is imprecisely - calculated... I should have done it, but it makes a point as it is :)
Registered Voters	Number of registered voters Why was this number to be entered by the presiding officer, yet it is in the records of IEBC? It should almost be the same at the polling stations located at the same precint
Actual Voted	Actual number of votes Should be nearly the same in polling stations located at same precincts
Turnout	Percentage of registered voters that voted. Should this be almost the same for polling station in the same location?
Originality	Is this the same image as was taken by the KIEMS kit? We retrieved the JPEG file from the PDF and subjected them to algorithm defined here: https://github.com/b3dk7/atomicjpeg Please read this How to interpret the values: 0 - 0.05 | 90% likely of being an original 0.05 - 0.1 | 70% likely of being an original 0.1 - 0.15 | 50% likely of bieng an original 0.15 - 0.2 | 30% likely of being an original 0.2 > infinity | 10% likely of being an original Most of Forms34 A were modified after being taken by the KIEMS kit!!
JPEG Profile	Was ICC_PROFILE available in the JPEG image within the PDF? Some of the images (about 13,000 out of the 46000) had this ICC_PROFILE record This shows us that, either, the PDFs / JPEGs were not created by the same app, or Some of the JPEGs were modified by a different app, not just the KIEMS kit alone It is this different app that caused the differences in the headers of JPEGs embedded in the PDFs