Beam particles interacting after a triggered event has satisfied the centrality requirement can fake a strangelet's signal if the second interaction occurs within the ADC's gates, and if energy is deposited at the correct position in the calorimeter.
At the trigger level this background can be sufficiently rejected with
a simple trigger that utilizes the multiplicity detector with thresholds set
much
lower than those used to select central interactions; some multiple-beam
interactions, however, will be recorded on tape. A conservative estimate of
this probability follows. The probability 
can be written as 
, where L is the fraction of triggered events with an unvetoed late
beam interaction, and F is the fraction of L events which deposit
``strangelet-class" energy in the ``right place" in the calorimeter.
where the three terms are the interaction rate (int/sec), gate length (sec), and estimated veto inefficiency, respectively.
The first term is the number of prompt protons and neutrons
in the experimental acceptance
per event for interactions which would fail to fire the low threshold
multiplicity requirement. These prompt particles are the only abundant
particles with
``strangelet-class" energies (in the range 5-20 GeV). The second bracketed
term is a geometrical factor which accounts for the overlap probability of the
showers of a prompt particle from the second event and a candidate
particle from the first (triggered) event. The showers are considered to
overlap if their centroids fall within 2.5 
of one another, where
the standard deviation of the shower centroid is taken from SPACAL data.
The probability of the entire process is therefore 
The product of the probability with the tracking system
rejection outlined in
Table 
is below the sensitivity planned for the experiment, and
so it should not pose problem in our analysis. There are
additionally other event characteristics that could be used to identify this
background if the need arises. First, there will be many charged tracks in
the spectrometer associated with the ``late" interaction. Most of these
tracks will be prompt, and they will all point back in time to an interaction
which followed the initial triggered interaction. Second, the mass measured
by momentum and time-of-flight will disagree with the mass measured by energy
and time-of-flight over large windows in TOF, since energy was artificially
added to the cluster in question. Third, the calorimeter cluster
time-of-flight would only fortuitously agree with the scintillator's timing
value since both interactions would be uncorrelated. Last, the calorimeter
cluster's time pattern may show the presence of two clusters, one from the
first and one from the second interaction.
A variation of this background results from contamination of the heavy ion beam with lighter ions or nucleons. In this situation the second beam interaction is caused by a low-A beam contaminant, and may be difficult for the multiplicity trigger to detect. At some level this will begin to affect E864's ability to achieve the desired sensitivity. Below is an estimate of the interaction rate of beam protons that will be tolerable.
The experiment plans to have a rejection of 
, so
it would be safe if this background occurred at the 
level.
Since the rejection of the tracking system is 
, the
multiple-beam background due to low-A beam protons would have to occur
at most every 
interactions.
Thus 
, where the variables have the same meaning
as above. Some of the numbers which make up L and F
will change because a single nucleon is responsible for the second interaction:
Therefore,
In other words, the experiment could conservatively tolerate as many
as 217,000 proton interactions per second before they would be a problem as
multiple-beam interactions. If our Au target is a 10% target for a Au beam,
then it is a 10% 
= 3.43% target for proton
collisions.
The 217,000 p-Au interactions would thus correspond to 
protons per second. This flux is 6/10 that of the Au beam itself.