PRELIMINARY RESULTS: Kassel_6.April.2006

PRELIMINARY RESULTS: Kassel_6.April.2006

A counter investigation into the murder of Halit Yozgat

Commissioned by the People’s Tribunal “Unravelling the NSU Complex”, Haus Der Kulturen Der Welt (HKW), Initiative 6 April, and submitted in evidence to the NSU trial, Higher Regional Court in Munich



1 Introduction

On the 6th of April 2006, Halit Yozgat, 21 years old, was murdered while attending the reception desk of the Internet café owned by his father, İsmail Yozgat, in Kassel, Germany. The murder was later attributed to a neo-Nazi group referred to as the National Socialist Underground (NSU). At the time of the killing, a secret service agent of the State Office for Constitutional Protection, (Landesamt für Verfassungsschutz) of the German state of Hessen named Andreas Temme (AT) was present in the café. In his interrogation by the police and in the subsequent NSU trial in Munich, AT denied being a witness to the incident. The court found that AT was present at the back room of the Internet café at the time of the murder and that from his position it was possible to have not witnessed the killing.

In November 2016 Forensic Architecture was commissioned by the organisers of the People’s Tribunal ‘Unravelling the NSU Complex’ to investigate this aspect of the case.

In order to undertake this analysis, Forensic Architecture constructed a life-size model of the Internet café and undertook a full reenactment of the incident. This took place at the House of World Cultures/HKW in Berlin between the 6th and 11th of March 2017.

The main question that this experiment came to address was: did Andreas Temme tell the truth about the incident?  Could he have witnessed the murder?

Witnessing in this context refers to a sensory contact with the incident. More precisely, the questions posed by this investigation were: could AT have heard the gunshots from his position in the back room, could he have seen the body as he left the café through the front room, and should he have smelled the residue of gun powder lingering in that room?

Several other questions arose in the process: did the police, the court and AT’s employers at the State Office for Constitutional Protection act in good faith when accepting his testimony and, if not, why?

Based on an examination of leaked police files, interviews with witnesses, spatial, aural, and olfactory reenactments and simulations, the project thus set out to examine not only the killing but also its possible cover-up and the protection of Temme from within the State Office for Constitutional Protection as a crime in its own right.



2 Methodology

We began by examining all available records — police reports and archives that included a police reenactment video, photographs from the crime scene, and computer and telephone logs — as well as plans, aerial and ground level images, and written and spoken testimonies. We examined transcripts of AT’s testimony in court. We have conducted our own measurements on site and interviewed witnesses.

Thereafter, we plotted out several possible scenarios across multiple timelines, identifying key characters, spaces, objects, and time frames. Within the physical model, we reenacted a number of different scenarios in order to determine the feasibility and plausibility of each timeline.

We also undertook three sensory tests, on which this brief preliminary report will concentrate:

2a Vision

The body of Halit Yozgat was first discovered by his father, İsmail Yozgat, when he returned to the café a few minutes after the murder. The father described the body of Halit as lying face down behind the reception desk. He produced a number of sketches (including some at the request of Forensic Architecture) depicting the body’s position. AT testified that he did not see the body when bending over to place a coin on the reception desk before leaving the café. AT performed a reenactment at the request of the Hessen State Police to support his testimony. A video of this reenactment was ultimately leaked and made public online. Aided by motion detection software and analogue measures we examined and modelled this reenactment video to establish the precise positions and movements of AT’s body, particularly of his head. We have thus recreated AT’s moving field of vision digitally within a computer model and with cameras (Go-Pro and digital camera using 33mm lens) attached to the head of an actor in a reenactment within the life-size model. This series of experiments were undertaken with the aim of investigating whether, even by AT’s own account, witnessing was possible.

2b Sound

The judges at the NSU Trial in Munich accepted that AT had been at the back room of the café, at a position known as PC-2, while the killing took place in the front room. As part of his testimony, AT, a trained marksman, claimed that he did not hear the two gunshots that killed Halit Yozgat.

Forensic Architecture contracted specialised weapons analysts, Armament Research Services (ARES), to record the sound signature of the weapon and ammunition used in the murder: a Česká CZ 83 pistol using 7.65mm Browning ammunition and a sound suppressor.

ARES sourced a Česká CZ 83 pistol and recorded 5 shots. They have verified that its sound signature and audio level were similar to another handgun of similar calibre — a Colt .32 pistol — using similar munitions. Both weapons offered equivalent peak sound signatures, all ranging from 157 to 158.5dBA.

The Colt .32 pistol was then alternately threaded with dry and wet sound suppressors and 5 shots were fired through each. None of these shots were suppressed below 130dBA.

Using both a digital simulation and life-size reenactment, Forensic Architecture together with a consultant from Anderson Acoustics, tested the audibility of these shots from AT’s position at PC-2.

For the purposes of this test, Forensic Architecture acquired a high-decibel active loudspeaker and located it at the position of the killer. We played the recorded gunshots sent by ARES within the life-sized space and in the computer model. We produced the level of the gunshots at 105dBA, some 25dBA lower than the sound of the shot. This was done under the assumption that if a shot at this level was heard, a louder shot certainly could be.

2c Smell

A gunshot in an interior space leaves a sharp smell of burnt gunpowder. When questioned by the German Federal Police in 2012, AT confirmed he was accustomed to handling guns and could identify the smell of gunpowder. However, he claimed that he smelled no such scent when moving from the back room of the café through to the front room towards the outside.

Forensic Architecture is working with a fluid dynamics specialist, Dr. Salvador Navarro-Martinez, Senior Lecturer at Imperial College, in order to calculate the dispersion and propagation of chemical components from the gunshots within space and time.  Dr. Navarro-Martinez calculated the volume of gases expected to be produced by such gunshots. We simulated this volume using smoke dispersers in order to visualise the dissipation and latency of smell within the space. Digital and analogue simulations were also used to model the fluid dynamics of ‘smell clouds’ in space and time. The findings from these tests were corroborated with a digital simulation that models the particle concentration, and therefore the perceptibility, of the smell of gunpowder.  We also measured the latency and potency of the smell in relation to the timeline constructed from witness testimonies and the police reenactment.



Initial Results

3a Vision

The reconstruction of AT’s moving field of vision establishes that the body of Halit Yozgat would have been visible to AT at the time he paused to place the coins on the reception desk.


3b. Sound

In the life-size model, the gunshot sound level was 86dBA at maximum level at AT’s seated position in PC-2, some 40dBA above the ambient sound level in the room and therefore audible.  86dB is the equivalent noise of a freight train at 15 metres away. This level of noise should be clearly audible over an ambient level of 40dBA, typical of living rooms, libraries, or small water streams.

In order to confirm that any additional sound paths present in the physical model would not significantly affect measured results and corroborate our findings, we created a computer simulation using ray tracing digital techniques.

Considering there was an open pathway for noise to travel through the open doorway between the two rooms in the café, it can be determined that the gunshot was audible from the position of AT.


The above graph shows the measured sound levels at AT’s position by PC-2 (red and orange) as well as the measured background level of noise in the space (blue). The gunshot should have been clearly audible.



3c Smell

The results of the olfactory experiment are still pending.



Photo Gallery


Forensic Architecture team

  • Eyal Weizman (Principle Investigator)
  • Christina Varvia   (Coordinator)
  • Stefanos Levidis
  • Omar Ferwati
  • Simone Rowat
  • Eeva Sarlin
  • Lawrence abu Hamdan (Advisor)


Collaborating Organizations

Comments are closed.