Conference PaperPDF Available

Opening the " Private Browsing " Data – Acquiring Evidence of Browsing Activities

Authors:

Abstract and Figures

The growing concern of users about the confidentiality of data generated by web browsing activities made browser developers include options for safer and confidential browsing in their products. For users those options, when functionally compliant with data security guidelines, guarantee online privacy. For law enforcement agents, this functionality introduces another obstacle for data acquisition towards evidence gathering. No matter which case, it is important to assess and validate private browsing techniques. The presented method shows that for some browsers it is possible to recover text and graphical data related to pages visited during private navigation, in clear violation of this tool basic functional requirement.
Content may be subject to copyright.
Opening the “Private Browsing” Data Acquiring Evidence of Browsing
Activities
Rodrigo de S. Ruiz 1, Fernando Pompeo Amatte, Kil Jin Brandini Park D. Sc. 3
1 Malware Analysis Nucleus (NUCAM)
Renato Archer Information Technology Center(CTI)
Campinas SP, Brazil.
2 Independent Researcher, Campinas SP, Brazil.
3 Computer Faculty (FACOM) Federal University of Uberlândia
Monte Carmelo MG, Brazil
rodrigosruiz@outlook.com, famatte@gmail.com, kil@facom.ufu.br
Abstract The growing concern of users about the
confidentiality of data generated by web browsing
activities made browser developers include options
for safer and confidential browsing in their
products.
For users those options, when functionally
compliant with data security guidelines, guarantee
online privacy. For law enforcement agents, this
functionality introduces another obstacle for data
acquisition towards evidence gathering.
No matter which case, it is important to
assess and validate private browsing techniques.
The presented method shows that for some
browsers it is possible to recover text and graphical
data related to pages visited during private
navigation, in clear violation of this tool basic
functional requirement.
Keywords: Private browsing, Browser safety,
Browser forensics.
1 INTRODUCTION
The growing concern of users with the
confidentiality of the data generated by the
activities developed in the course of navigation
through web pages fostered the development of
navigation options that offer greater degree of
security and confidentiality of the data.
The promise of the developers regarding the
operation of this feature is to prevent others to
reconstruct the steps the user took during his
online activities.
On Mozilla’s page we found this commercial
text about privacy and private browsing:
“Sometimes it’s nice to go undercover:
Open a private window and protect your
browsing history. You can switch
between private and normal windows
quickly, so it’s easy to go back to
what you were doing before. This
feature is great if you’re doing your
online banking on a shared computer or
checking email from an Internet café.”
[1]
On Chrome’s browser, when the user enable
incognito mode, the new tab opened displays
the following message:
“You came in incognito mode. Pages you
view in this window will not appear in
your browser history or search history
will not leave other traces, like
cookies, on your computer after you
close all incognito windows open.
However, all the downloads you make or
bookmarks you create will be
preserved.”
On Safari’s page we found this commercial
text about privacy and private browsing:
“...Safari can keep your browsing
history private. When you turn on
Private Browsing, Safari does not
remember the pages you visited, your
search history, or your AutoFill
information...”[2]
On IE’s page we found this commercial text
about privacy and private browsing:
“While you are surfing the web using
InPrivate Browsing, Internet Explorer
stores some informationsuch as
cookies and temporary Internet files
so the webpages you visit will work
correctly. However, at the end of your
InPrivate Browsing session, this
information is discarded...”[3]
Proceedings of the International Conference on Information Security and Cyber Forensics, Kuala Terengganu, Malaysia, 2014
ISBN: 978-1-941968-01-7 ©2014 SDIWC
72
On the one hand such a feature, if operating
perfectly aligned with security guidelines,
provides the user privacy in their online
activities, on the other hand it is clear that in
case of unlawful behavior, law enforcement
officers have to deal with this layer of protection
to obtain the necessary data to provide evidence
during the course of an investigation.
In both cases, it is important to verify the
actual functionality of such a feature, if
available implementations actually provide the
degree of confidentiality offered, or if there are
flaws that allow the retrieval of online activity
data.
This paper is an extended version of a work
previously presented by the authors [4], with
additional results and analysis, and is structured
in the following topics:
Method and Tests, which presents the
method applied to tests performed in various
browsers with the private browsing feature
enabled.
Results and Discussion, which presents the
results obtained by the tests adopted and
discusses these results.
Finally, follow the conclusions, further
studies and references used.
2 METHOD AND TESTS
When testing a security feature, it is
necessary to define its functional requirements
and the profile of the attacker who will try to
disable or override this feature.
In a paper on the analysis of private browsing
functionality, [5] lists the profiles of potential
attackers, security models to be checked and the
objectives to be met by browsers that implement
private browsing. In this work, we start from the
methodological framework presented by [5], for
the construction of the following
methodological model:
The profile of the attacker considered
assumes that he has local access to the user
machine. Consequently, attempts to circumvent
the system of private browsing will occur from
an image taken from the user's machine hard
drive.
As the focus of the evaluation is the private
browsing feature, we considered that the user
does not adopt other security tools or techniques
that could exert influence on the access of the
data generated during navigation. Thus, we did
not conduct any test with the adoption of
cryptographic methods in the disk of the user's
machine.
Furthermore, this paper focus on searching
the user's machine for fragments of data from
which text or images that brings information
about pages visited could be extracted.
Therefore, the specific analysis of changes to
files used by browsers such as history, cookies,
cache and certificates was not performed. Such
analysis can be found in [5] and [6].
We tested Internet Explorer browser on bare
metal hardware with the use of four notebooks
equipped with Windows 7 Pro SP1.
For the other tests performed, we created a
standard guest virtual machine - with the
operating system Windows 7 Pro - in the host
operating system - Windows 7 Pro - using the
virtualization software Virtual Box [7].
An export (snapshot) of the newly installed
Windows machine was created, considering the
possible need for future comparison of the base
guest machine with guest machines running the
different browsers tested.
The browsers tested were Internet Explorer
10, Firefox 24.0_1, Google Chrome
30.0.159969M_1 and Safari 5.1.7_1. The base
guest virtual machine for each browser was
replicated 4 times, each to be used in the four
different tests performed on each browser.
Based on those configurations, four different
tests for each browser in private browsing mode
were applied:
Test S (Shutdown): Consists of visiting a
web site available on the internet, making
operations to interact with the site, finish the
execution of the browser correctly and
generating the virtual machine image for
analysis.
Test F (Freeze): Consists of visiting a web
site available on the Internet, making operations
to interact with the site and with the browser
still active, generating the virtual machine image
for analysis.
Proceedings of the International Conference on Information Security and Cyber Forensics, Kuala Terengganu, Malaysia, 2014
ISBN: 978-1-941968-01-7 ©2014 SDIWC
73
Test K (Kill process): Consists of visiting a
web site available on the internet, making
operations to interact with the site, requesting
that the operating system interrupt the browser
execution and generating the virtual machine
image for analysis.
Test P (Power down): Consists of visiting a
web site available on the internet, making
operations to interact with the site, requesting
the virtualizer to turn off the virtual machine -
simulating a power outage - generating the
virtual machine image for analysis.
For each test performed, the virtual machine
image generated will be analyzed through the
application of the program strings [8] found in
many different Linux distributions.
This program is used for the search of strings
inside the virtual machine images that could
present relation to the webpage visited.
The images of the virtual machines will be
analyzed for the search of graphic files
associated with the visited webpage, through the
usage of the foremost program [9], a renowned
forensic tool for extraction of files - "data
carving" - of different formats.
This tool works as follows: It reads a block
of data - memory, disk or files - and looks for
signatures related to files of well-known
formats. It is noteworthy that in the present
research we investigated only the persistent
memory (i.e. physical and virtual disk).
Since these signatures are a sequence of
bytes, there is the chance of occurrence of false
positives and therefore the capture of incorrect
file.
Furthermore, it is important to note that there
exist several known problems associated with
the use of tools aiming for "data carving”, for
example, limitations to the treatment of non-
contiguous data. Thus, it is possible that an
image whose sequence of bytes is dispersed will
not be fully recovered, despite its possible
existence in the block of data analyzed.
The WinHex tool was also used to search for
keywords found in the navigated webpage.
3 RESULTS
Aiming to simulate an actual visit to any
website available on the internet, a random
selection was made, and the site chosen for the
experiment was the [10]. Since some site
information is proprietary, the figures recovered
during the test will be only partially reproduced
in the present work. We would like to
acknowledge that those information are
copyright of their respective owners.
SAFARI Browser
For the Safari browser, the following results
were obtained:
F test (freeze)
Figure 1- storage.discovery.com string located in
virtual machine´s image.
No image fragments were found on the
virtual machine´s hard disk image.
K Test (kill process)
Figure 2- “discovery.com string located in virtual
machine´s image.
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 3 Image recovered on hard disk image
analysis and found on Discovery.com website.
The strings utility could also recover text
references in the virtual machine hard disk
image analysis that indicated the webpage was
visited:
p://dsc.discovery.com/videos
Proceedings of the International Conference on Information Security and Cyber Forensics, Kuala Terengganu, Malaysia, 2014
ISBN: 978-1-941968-01-7 ©2014 SDIWC
74
http://store.discovery.com/?ecid=PRF-DSC-
101345&pa=PRF-DSC-101345
P Test (Power down)
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 4 Image recovered on hard disk image
analysis and found on Discovery.com website.
The strings utility could also recover text
references in the virtual machine hard disk
image analysis that indicated the webpage was
visited:
http://store.discovery.com/discovery/layout/favicon.ico
http://dsc.discovery.com/
http://games.dsc.discovery.com/
http://dsc.discovery.com/tv-shows
http://store.discovery.com/discovery/layout/favicon.ico
S Test (Shutdown)
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 5 Image recovered on hard disk image
analysis and found on Discovery.com website.
The strings utility could also recover text
references in the virtual machine hard disk
image analysis that indicated the webpage was
visited:
http://dsc.discovery.com/tv-shows
http://dsc.discovery.com/
http://store.discovery.com/discovery/layout/favicon.ico
http://dsc.discovery.com/videos
america.discovery.com.edgesuite.net
velocity.discovery.com
metrics.discovery.com
orate.discovery.com
animal.discovery.com.edgesuite.net
The results obtained for the Safari browser
tests are grouped in table 1:
Table 1 Results for Safari Browser
F Test
K Test
P Test
S Test
Page address
recover
Yes
Yes
Yes
Yes
Picture recover
No
Yes
Yes
Yes
FIREFOX browser:
F Test (freeze)
Figure 6 –“sc.discovery.com/video-topics string
located in the virtual machine´s image.
No images related to the webpage visited
were found on the virtual machine hard disk
image analysis.
K Test (kill process)
Figure 7 “discovery.com string located in the
virtual machine´s image.
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 8 Image recovered on hard disk image
analysis and found on Discovery.com website.
The strings utility could also recover text
references in the virtual machine hard disk
image analysis that indicated the webpage was
visited:
Proceedings of the International Conference on Information Security and Cyber Forensics, Kuala Terengganu, Malaysia, 2014
ISBN: 978-1-941968-01-7 ©2014 SDIWC
75
C:\Program Files\Mozilla Firefox\firefox.exe
ttp://games.dsc.discovery.com/
/ttp://dsc.discovery.com/videos
http://games.dsc.discovery.com/word-games
http://games.dsc.discovery.com/sport-games
https://securestore.discovery.com/cart.php
https://securestore.discovery.com/cart.php
store.discovery.com
http://games.dsc.d
P Test (Power down)
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 9 Image recovered on hard disk image
analysis and found on Discovery.com website.
The strings utility could also recover text
references in the virtual machine hard disk
image analysis that indicates the webpage was
visited:
investigation.discovery.com.edgesuite.net
netstorage.discovery.com.edgesuite.net
netstorage.discovery.com
netstorage.discovery.com.edgesuite.net
netstorage.discovery.com.edgesuite.net
netstorage.discovery.com.edgesuite.net
games.dsc.discovery.com
S Test (Shutdown)
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 10 Image recovered on hard disk image
analysis and found on Discovery.com website.
The strings utility could also recover text
references in the virtual machine hard disk
image analysis that indicated the webpage was
visited. A fraction of strings retrieved in this test
follows:
Fdsc.discovery.com%2Fvideo-
topics%2Fadventure&u=oeu1381760545360r0.43558277
74372748&wxhr=true&t=1381760579498&f=34093708
6
http://dsc.discovery.com/
h;e++)if(a[e].name=="keywords")if(b=="")b=a[e].c
ontent;else b+=", "+a[e].content;else
if(a[e].name=="description")c=a[e].content;if(!(b.length
+c.length>eb)){z("dmk",b);z("dmd",c)}}function ub(){var
a="__cmb",b=[];for(var c in
aa)c.indexOf(a)==0&&b.push(c
s_sess=%20s_cc%3Dtrue%3B%20s_campaign%3DPRF-
DSC-101345%3B%20s_sq
Table 2 Results for Fire Fox Browser
F Test
K Test
P Test
S Test
Page address
recover
Yes
Yes
Yes
Yes
Picture recover
No
Yes
Yes
Yes
GOOGLE CHROME Browser
F Test (freeze)
Some strings related to the webpage were
also found in hard disk analysis:
Figure 11- “http://dsc.discovery.com string
located in the virtual machine´s image.
Figure 12- “discovery.com< /domain>” string
located in the virtual machine´s image.
K Test (kill process)
Proceedings of the International Conference on Information Security and Cyber Forensics, Kuala Terengganu, Malaysia, 2014
ISBN: 978-1-941968-01-7 ©2014 SDIWC
76
Figure 13 “discovery.com/tv-shows” string
located in the virtual machine´s image.
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 14 Image recovered on hard disk image
analysis and found on Discovery.com website.
Figure 15 Image recovered on hard disk image
analysis and found on Discovery.com website.
P Test (Power down)
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 16 Image recovered on hard disk image
analysis and found on Discovery.com website.
The strings utility could also recover text
references in the virtual machine hard disk
image analysis that indicated the webpage was
visited. A fraction of strings retrieved in this test
follows:
//dsc.discovery.com/
://static.ak.facebook.com/connect/xd_arbiter.php?vers
ion=27#cb=fdde13148&domain=dsc.discovery.com&ori
gin=http%3A%2F%2Fdsc.discovery.com%2Ff2a7e0cd34
&relation=parent&error=unknown_user
/dsc.discovery.com/tv-shows
://dsc.discovery.com/
://dsc.discovery.com/
://dsc.discovery.com/
http://dsc.discovery.com/tv-shows
http://dsc.discovery.com/tv-shows
http://dsc.discovery.com/
S Test (Shutdonw)
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 17 - Image recovered on hard disk image
analysis and found on Discovery.com website.
The strings utility could also recover text
references in the virtual machine hard disk
image analysis that indicated the webpage was
visited. A fraction of strings retrieved in this test
follows:
":"Survivorman Videos","srtUrl":"","uuid":"8e18dcd9-
8d1d-11e2-a7b7-06a90ff35868","bdat":"must
watch","keywords":"survivorman,10 days,ten,days,must
watch,mexico,tiburon,deserted,island,les
stroud,survival,survivor,man,water,pool,algae,fresh,cane
,reed,sludge","mediaType":"lift","mp4":[{"bitrate":"110
k","src":"http://discsmil.edgesuite.net/digmed/hdnet/07/a
7/13776400801197_102MissingPiece-
110k.mp4"}f.akamaihd.net/i/digmed/hdnet/98/9a/137764
01201197_104Stove-
,400k,110k,200k,600k,800k,1500k,3500k,.mp4.csmil/mast
er.m3u8","networkId":"DSC","thumbnailURL":"http://ne
tstorage.discovery.com/feeds/brightcove/asset-
thumbnails/dsc/0a5dbdfa893fec1f556a7d81c5b28bc470e
cbb0e_0a5dbdfa893fec1f556a7d81c5b28bc470ecbb0e.jp
g"
Table 3 Results for Chrome Browser
F Test
K Test
P Test
S Test
Page address
recover
Yes
Yes
Yes
Yes
Picture recover
No
Yes
Yes
Yes
INTERNET EXPLORER Browser
Proceedings of the International Conference on Information Security and Cyber Forensics, Kuala Terengganu, Malaysia, 2014
ISBN: 978-1-941968-01-7 ©2014 SDIWC
77
F Test (freeze)
Figure 18 “discovery.com string located in the
virtual machine´s image.
No images related to the webpage visited
were found on the virtual machine hard disk
image analysis.
K Test(kill process)
Figure 19 “http://store.discovery.com/js/ajax/
string located in the virtual machine´s image.
No images related to the webpage visited
were found on the virtual machine hard disk
image analysis.
P Test (Power down)
Images related to the webpage visited were
found on the virtual machine hard disk image
analysis:
Figure 20 Image recovered on hard disk image
analysis and found on Discovery.com website.
S Test (Shutdown)
On this test, another step taken was the
analysis of log files generated by the Internet
Explorer browser. It is easy to see that the page
address is easily visible inside a log file:
Figure 21 Log file found using only the explorer
and notepad. They demonstrate the system failure
(string http://dsc.discovery.com found) in the private-
IE10.
Table 4 Results for IE10
F Test
K Test
P Test
S Test
Yes
Yes
No
Yes
No
No
Yes
No
Further analysis to prospect the files and
directories involved in the data leakage
generated the following results:
In all browsers, some of the data associated
with the navigation could be extracted from the
file pagefile.sys. This proves that part of the
data is leaking through the paging process´s
storage mechanism used by the operating
system.
In Internet Explorer´s case, more data could
be found in a file located at the directory:
\user\<username>\appdata\local\microsoft\wi
ndows\temporary internet
files\low\content.ie5\ndm4l4gv\
On Chrome´s case, more data could be found
in the file:
\user\administrador\appdata\local\microsoft\
windows\webcache\webcachev01.dat
Those files points to the fact that navigation
data is leaking from cache files used by the
browsers.
Proceedings of the International Conference on Information Security and Cyber Forensics, Kuala Terengganu, Malaysia, 2014
ISBN: 978-1-941968-01-7 ©2014 SDIWC
78
4 DISCUSSION
From the data generated by the tests, it is
possible to assume that every implementation of
the private browsing functionality in all
browsers tested demonstrate some type of
failure.
In some cases, those flaws allow an attacker
to identify the pages visited by the user. In other
cases, they generate enough data to allow the
partial reconstruction of the pages visited.
We contact the developers about the results
and obtained some mixed comments.
Microsoft answer to our request for
comment:
“...We do encourage security
researchers we are working with to
present their research at events...
...The issue is still being scoped
and researched. I will let you know
once that has finished and a servicing
decision has been made”
From the information about the private
browsing functionality and the answers
received, it is possible to extract that the
average user is not well informed of the
limitations inherent to the implementations of
the service.
5 CONCLUSION
In all four types of tests performed, it is
possible to verify that all browsers tested
presented flaws in their private browsing
feature.
Those flaws generates data that remains
available in the system and allow not only the
identification of pages visited but in some cases
also to partially rebuild them.
Browsers promises to leave no traces of the
navigation activities of users. This work proves
that privacy as advertised is not provided.
In face of the results obtained, we would like
to recommend the developers to explicitly alert
the users about the limitations of the private
browsing functionality implementation.
We would like to praise Microsoft´s answer
because they both acknowledged the
information received and approved the release
of the study.
If on one hand this is a negative point for the
user, on the other hand those flaws facilitate the
work of law enforcers in cases where there is
need for the data related to the navigation
activity.
6 FURTHER STUDIES
In future researches, we plan to analyze the
mechanisms and data structures - both browser
and operating system related - involved in the
browsing activities data leakage in-depth. This
line of study could bring forth new techniques
to avoid the problems presented in this paper on
the implementations of the private browsing
functions.
7 REFERENCES
[1] Mozilla private browsing. Available at:
http://www.mozilla.org/en-US/firefox/features/ Accessed
at: Oct, 24, 2013
[2] Archived - Mac Basics: Safari 5.1. Available at:
http://support.apple.com/kb/ht4550. Accessed at: Nov,
21, 2013.
[3] What is InPrivate Browsing? Available at:
http://windows.microsoft.com/en-us/windows7/what-is-
inprivate-browsing. Accessed at: Nov, 21, 2013.
[4] RUIZ, R. S., AMATTE, F. P., PARK, K. J. B.
Tornando Pública a Navegação “InPrivate”.
Prooceedings of the IcoFCS2012. Available at:
http://www.icofcs.org/2012/ICoFCS2012_Full.pdf.
Accessed at: Nov, 21, 2013.
[5] AGGARVAL, G. BURSZTEIN, E. JACKSON, C.
BONEH, An Analysis of Private Browsing Modes in
Modern Browsers. USENIX 2010, Available at:
http://crypto.stanford.edu/~dabo/pubs/papers/privatebrow
sing.pdf. Accessed at: Jun, 30, 2012.
[6] MAHENDRAKAR, A. IRVING, J. PATEL, S.
Forensic Analysis of Private Browsing Mode in Popular
Browsers. Available at: http://mocktest.net/paper.pdf.
Accessed at: Jun, 301, 2012.
[7] VirtualBox tool. Available at:
http://www.oracle.com/technetwork/server-
storage/virtualbox/downloads/index.html. Accessed at:
Jun, 13, 2013.
[8] Strings man page. Available at:
http://linux.die.net/man/1/strings. Accessed at Jun, 30
2012.
[9] Foremost website. Available at:
http://foremost.sourceforge.net/ Access at: Oct 14, 2012.
[10] Discovery.com website. Available at:
http://dsc.discovery.com. Accessed at: Oct, 07, 2012.
Proceedings of the International Conference on Information Security and Cyber Forensics, Kuala Terengganu, Malaysia, 2014
ISBN: 978-1-941968-01-7 ©2014 SDIWC
79
... In a study about the private browsing feature, Ruiz et al. adopted the second line of methodology. 11 Applying signatures and string matching, the authors were able to extract data related to visited websites in various browsers even though the private browsing functionality was in use. ...
Article
Full-text available
E-banking systems rely on browsers and other financial applications to provide a secure service. As this paper shows, however, many e-banking systems fail to provide the requisite level of confidentiality. This paper focuses on vulnerabilities related to the leakage of login information from the client side. It demonstrates that by using forensic techniques and tools, it was possible to acquire login-related data from a number of websites. This proves that important authentication data used in navigation activities remain on disk, posing a clear threat to confidentiality. https://www.henrystewartpublications.com/jpss/v11
Conference Paper
Today the Internet has revolutionized it into a cyberspace, mainly due to its attributes of convenience, interoperability, fast search and confidence. It provides an affordable and secure way to spontaneously link people and computers across geographical boundaries. The massive economy of online services and digital information has dramatically changed our life today. Digital private information is sensitive personal information digitally stored in cyberspace, or being transmitted through the Internet. This may include gender, age, race, location, and records on health, finance, and even online activities. Although these types of data are usually privately stored, they can be intentionally or unintentionally revealed while surfing the Internet. To address the digital privacy concern, current Web Browsers developed the Private Browsing Mode (PBM) which promises to serve the user to surf the internet without leaving traces on the local machine. In this paper, we study the Internet privacy using Private Browsing Mode (PBM) with four test scenarios on several popular browsers. The test results indicate that, with current design of web sites and web browsers, even under PBM, privacy information could still be retrieved. We hope the finding can provide us an opportunity to improve future design of PBM in browsers.
Article
Private browsing refers to a specialized mode com-mon in current mainstream internet browsers which is designed to leave no traces of browsing activity performed while in the mode. In this paper we ex-amine various internet browsers in order to deter-mine what traces of browsing activity remains in physical memory after using the private browsing modes of each browser. We also analyze some of the mechanisms used by private browsing modes to prevent leaving traces. We then propose counter-measures to some of these methods.
Conference Paper
We study the security and privacy of private browsing modes recently added to all major browsers. We first pro- pose a clean definition of the goals of private browsing and survey its implementation in different browsers. We conduct a measurement study to determine how often it is used and on what categories of sites. Our results suggest that private browsing is used differently from how it is marketed. We then describe an automated technique for testing the security of private browsing modes and report on a few weaknesses found in the Firefox browser. Fi- nally, we show that many popular browser extensions and plugins undermine the security of private browsing. We propose and experiment with a workable policy that lets users safely run extensions in private browsing mode.
Safari 5.1. Available at: http://support.apple.com/kb/ht4550
  • Archived-Mac Basics
Archived -Mac Basics: Safari 5.1. Available at: http://support.apple.com/kb/ht4550. Accessed at: Nov, 21, 2013.