You look at the histograms. You print the histograms on heavy white paper. The first histogram shows a peak at the center. The second histogram shows a peak to the left. The third histogram shows a peak to the right. These three instruments are the same model. These three instruments use the same parts. The results on the paper are not the same.
You feel the weight of your shoe in your hand. You just killed a spider on the floor. The spider is a smear now. The spider had no warning. The shoe was a fact of the environment. The problem with the histograms is also a fact of the environment. You did not choose this fact.
Figure 1: Observed variance across three identical instrument models using identical parts.
You find the datasheet for the flow cell. The datasheet is blue. The datasheet lists the material of the cell. The material is quartz. The datasheet lists the window size of the cell. The window size is 5.2 millimeters. You look for the channel tolerance. The channel is the hole in the center. The fluid goes through the channel. The particles go through the channel.
You find the line for channel tolerance on the datasheet. There is no number on the line. The line is empty.
The Hidden Geometry of Fluid Velocity
The channel width dictates the flow velocity. The flow velocity dictates the particle transit time. If the channel is wide, the fluid moves slow. If the channel is narrow, the fluid moves fast. You did not choose the width of the channel. The vendor chose the width of the channel. The vendor did not tell you the choice.
You realize the histograms are different because the channels are different. One channel is 252 microns wide. Another channel is 238 microns wide. The vendor sold both channels as the same part number. The vendor did not commit to a number. The vendor committed to a process. The process is cheap. The process is variable.
You bought a tolerance you never read. Your readings inherited this tolerance. Your data is the child of a manufacturing shortcut. You perceive the error now. The error was always there. The error was in the blank space on the datasheet.
“The impact is the only thing that does not lie. When a car hits a wall, the metal bends. The metal bends because of the geometry. If the geometry changes by a millimeter, the bend changes.”
– Ian J.P., Car Crash Test Coordinator
Ian J.P. knows about force. Ian J.P. knows about variables. The car crash is a measurement. Your flow cytometer is a measurement. The flow cell is the wall. The particles are the cars.
The sheath fluid surrounds the sample. The sample is a thin stream. The sample stays in the center. This is hydrodynamic focusing. The geometry of the channel creates the focus. The channel walls must be flat. The channel walls must be parallel. If the channel walls are uneven, the focus shifts.
The detector looks at the center of the window. The particles are no longer in the center of the window. The detector sees noise. The detector sees shadows. The detector sees nothing.
You call the vendor. You ask for the channel tolerance. The salesperson does not have the number. The salesperson says the part meets the specification. You tell the salesperson the specification is empty. The salesperson says the process is proprietary.
Proprietary is a word for a secret. The secret is that the vendor does not measure the channel. The vendor grinds the glass. The vendor polishes the glass. The vendor hopes the glass is correct. Hope is not a measurement. Hope is not an engineering value.
You cannot calibrate away a physical geometry error. The fluid dynamics follow the glass. The glass is the law. If the law is hidden, you cannot follow the law.
Accountability as a Metric
HookeLab makes the cells differently. HookeLab measures the channels. The channel tolerance is . HookeLab writes the number on the paper. The number is a commitment. The number is a promise.
When the number is on the paper, the vendor is accountable. When the number is missing, the buyer is responsible for the failure. You look at the spider on the floor. The spider was living its life. The shoe changed the life of the spider. The shoe was long. The shoe had a specific mass. The shoe had a specific velocity. The spider died because of the physics of the shoe.
Your instrument dies because of the physics of the channel. The channel is a small shoe. The channel hits your data every .
The engineer at the next desk asks about the histograms. The engineer wants to know why the peaks drift. You show the engineer the datasheet. You point to the empty line. The engineer looks at the empty line. The engineer does not speak. The engineer recognizes the silence. The vendor saved money on metrology. You spent money on a mystery.
A custom flow cell requires precision. The window alignment must be exact. The window alignment affects the laser path. If the window is tilted by 0.5 degrees, the laser refracts. The refraction creates a ghost image. The ghost image looks like a particle. The ghost image is a lie. You cannot distinguish the lie from the truth if the geometry is unknown.
You think about the production line. You think about 114 instruments. Each instrument has a flow cell. If the tolerance is not specified, each instrument is a different machine. You are not building a fleet of instruments. You are building 114 individual problems. You will spend your Saturdays fixing the problems. You will spend your Sundays wondering why the problems exist. The answer is in the blue datasheet. The answer is the missing number.
Pits, Cracks, and Turbulence
The fluid moves in layers. This is laminar flow. The layers are delicate. If the channel wall has a pit, the layer breaks. The break creates turbulence. Turbulence is the enemy of the signal. The vendor does not check for pits. The vendor checks for cracks. A pit is not a crack. A pit is just a variation. The variation is yours now. You own the pit. You own the turbulence. You own the histogram that looks like a mountain range.
Precision is expensive. Accountability is more expensive. A vendor who leaves a spec blank is a vendor who is afraid of the truth. The truth is that their process cannot hold a tight tolerance. They sell you the variance as if it were a law of nature. It is not a law of nature. It is a limitation of the factory.
You pick up the shoe. You put the shoe on your foot. You walk to the trash can. You throw the white paper with the histograms in the trash. The paper is heavy. The paper hits the bottom of the can. You will call a new vendor. You will ask for a number. You will not accept a blank line. You will not accept a secret.
The sun is bright outside. The light comes through the window. The light hits the flow cell on your desk. The light bends through the quartz. You see the rainbow on the wall. The rainbow is beautiful. The rainbow is also a measurement of the refractive index.
If the quartz is low quality, the rainbow is fuzzy. If the quartz is high quality, the rainbow is sharp. Everything is a measurement. Everything has a tolerance. If you do not know the tolerance, you do not know the part. You only know the price. The price is the least important thing about the part.
You open a new file on the computer. You type the requirements for the new flow cell. You type “Channel Tolerance: ±0.02 mm.” You type “Material: JGS-1 Quartz.” You type “Window Alignment: < 10 microns." You are writing a new law. You are building a new environment. The next spider will not be a surprise. The next histogram will be a peak. The peak will stay in the center. The center is where the truth lives.
You grasp the difference now. The difference is the number. The number is the only thing that matters in the dark.